* [PATCH v5 0/4] GDB support for DW_AT_trampoline
@ 2023-08-16 21:53 Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 1/4] gdb, dwarf: add support for DW_AT_trampoline in DWARF reader Abdul Basit Ijaz
` (5 more replies)
0 siblings, 6 replies; 13+ messages in thread
From: Abdul Basit Ijaz @ 2023-08-16 21:53 UTC (permalink / raw)
To: gdb-patches; +Cc: abdul.b.ijaz, JiniSusan.George, tom, eliz, blarsen
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com>
Hi,
please find attached a series of patches that attempts to enable GDB to
handle the DW_AT_trampoline attribute on functions and inlined
functions.
DW_AT_trampoline can be emitted by the compiler for functions that are
compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of
DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the
target for a given function. Patch 3 adapts GDB's stepping behavior
when dealing with trampolines and attempts to - by default - hide these
trampolines from the user. When about to step into a trampoline,
instead, GDB will try and step through the trampoline and directly
towards the target. Patch 4 in V2 now skip trampoline frames
in the stack for printing or finish command. A new setting has been
introduced to turn off this modified stepping behavior or printing of stack
for trampolines.
The motivation for these patches comes from ifx which emits the
attribute for some of its compiler generated functions. As I do not
know of any other compiler (especially gcc/gfortran) emitting
DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to
test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-)
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v5 1/4] gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
@ 2023-08-16 21:53 ` Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 2/4] gdb/symtab: add lookup for trampoline functions Abdul Basit Ijaz
` (4 subsequent siblings)
5 siblings, 0 replies; 13+ messages in thread
From: Abdul Basit Ijaz @ 2023-08-16 21:53 UTC (permalink / raw)
To: gdb-patches; +Cc: abdul.b.ijaz, JiniSusan.George, tom, eliz, blarsen
DW_AT_trampoline can be used to describe compiler generated functions
that serve some intermediary purpose on making a call to another
function. A compiler can emit this tag in order to help a debugger hide
the trampolines from a user.
The attribute is only applicable to DW_TAG_subroutine and
DW_TAG_inlined_subroutine tags. It contains information about the
trampoline target either as a reference to its DIE, as its address or
its name. DW_AT_trampoline can also be a flag indicating that the DIE
is a trampoline or not without specifying the target (e.g. if it is
unknown).
This patch adds support to GDB for reading the DW_AT_trampoline
attribute. It stores the attribute and its value in the type_specific
part of a GDB type. This patch is implemented in preparation of the
following patches, which will add a mechanism to hide DW_AT_trampoline
subroutines from the user.
2023-08-16 Nils-Christian Kempke <nils-christian.kempke@intel.com>
---
gdb/dwarf2/read.c | 46 ++++++++++++++++++-
gdb/gdbtypes.c | 34 +++++++++++++-
gdb/gdbtypes.h | 113 ++++++++++++++++++++++++++++++++++++++++++++--
3 files changed, 186 insertions(+), 7 deletions(-)
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index a64f82bd24a..748cdbc4eb2 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -14656,6 +14656,50 @@ read_subroutine_type (struct die_info *die, struct dwarf2_cu *cu)
if (prototyped_function_p (die, cu))
ftype->set_is_prototyped (true);
+ /* If this is a trampoline function store it and its target here. */
+ attr = dwarf2_attr (die, DW_AT_trampoline, cu);
+ if (attr != nullptr)
+ {
+ TYPE_FUNC_FLAGS (ftype) |= FUNC_TYPE_TRAMPOLINE;
+ TYPE_TRAMPOLINE_TARGET (ftype)
+ = (trampoline_target *) TYPE_ZALLOC (ftype,
+ sizeof (trampoline_target));
+
+ /* A DW_AT_trampoline can be either an address, a flag, a reference or a
+ string. */
+ if (attr->form_is_string ())
+ TYPE_TRAMPOLINE_TARGET (ftype)->set_target_name
+ (attr->as_string ());
+ else if (attr->form_is_ref ())
+ {
+ die_info *target_die;
+ dwarf2_cu *target_cu = cu;
+ unrelocated_addr lowpc;
+
+ target_die = follow_die_ref (die, attr, &target_cu);
+
+ if (dwarf2_get_pc_bounds (target_die, &lowpc, NULL, target_cu,
+ nullptr, nullptr) <= PC_BOUNDS_INVALID)
+ complaint (_("DW_AT_trampoline target DIE has invalid "
+ "low pc, for referencing DIE %s[in module %s]"),
+ sect_offset_str (die->sect_off),
+ objfile_name (objfile));
+ else
+ {
+ lowpc = cu->per_objfile->adjust (lowpc);
+ TYPE_TRAMPOLINE_TARGET (ftype)->set_target_addr (lowpc);
+ }
+ }
+ else if (attr->form_is_unsigned ())
+ TYPE_TRAMPOLINE_TARGET (ftype)->set_target_flag (attr->as_boolean ());
+ else
+ {
+ unrelocated_addr target_addr = attr->as_address ();
+ target_addr = cu->per_objfile->adjust (target_addr);
+ TYPE_TRAMPOLINE_TARGET (ftype)->set_target_addr (target_addr);
+ }
+ }
+
/* Store the calling convention in the type if it's available in
the subroutine die. Otherwise set the calling convention to
the default value DW_CC_normal. */
@@ -14673,7 +14717,7 @@ read_subroutine_type (struct die_info *die, struct dwarf2_cu *cu)
if the DWARF producer set that information. */
attr = dwarf2_attr (die, DW_AT_noreturn, cu);
if (attr && attr->as_boolean ())
- TYPE_NO_RETURN (ftype) = 1;
+ TYPE_FUNC_FLAGS (ftype) |= FUNC_TYPE_NO_RETURN;
/* We need to add the subroutine type to the die immediately so
we don't infinitely recurse when dealing with parameters
diff --git a/gdb/gdbtypes.c b/gdb/gdbtypes.c
index d7db7beb554..096c5c85a75 100644
--- a/gdb/gdbtypes.c
+++ b/gdb/gdbtypes.c
@@ -5091,6 +5091,33 @@ print_fixed_point_type_info (struct type *type, int spaces)
type->fixed_point_scaling_factor ().str ().c_str ());
}
+/* Print the contents of the TYPE's self_trampoline_target, assuming that its
+ type-specific kind is TYPE_SPECIFIC_FUNC and is_trampoline is not 0. */
+static void
+print_trampoline_target_info (struct type *type, int spaces)
+{
+ switch (TYPE_TRAMPOLINE_TARGET (type)->target_kind ())
+ {
+ case TRAMPOLINE_TARGET_ADDR:
+ gdb_printf ("%*starget addr: 0x%s\n", spaces + 2, "",
+ print_core_address (type->arch_owner (),
+ (CORE_ADDR) TYPE_TRAMPOLINE_TARGET (type)
+ ->target_addr ()));
+ break;
+ case TRAMPOLINE_TARGET_NAME:
+ gdb_printf ("%*starget name: %s\n", spaces + 2, "",
+ TYPE_TRAMPOLINE_TARGET (type)->target_name ());
+ break;
+ case TRAMPOLINE_TARGET_FLAG:
+ gdb_printf ("%*starget flag: %d\n", spaces + 2, "",
+ TYPE_TRAMPOLINE_TARGET (type)->target_flag ());
+ break;
+ default:
+ gdb_assert_not_reached ("unhandled trampoline target kind");
+ break;
+ }
+}
+
static struct obstack dont_print_type_obstack;
/* Print the dynamic_prop PROP. */
@@ -5417,6 +5444,10 @@ recursive_dump_type (struct type *type, int spaces)
gdb_printf ("%*scalling_convention %d\n", spaces, "",
TYPE_CALLING_CONVENTION (type));
/* tail_call_list is not printed. */
+ gdb_printf ("%*sfunc_type_flags 0x%x\n", spaces, "",
+ (unsigned int) TYPE_FUNC_FLAGS (type));
+ if (TYPE_IS_TRAMPOLINE (type))
+ print_trampoline_target_info (type, spaces);
break;
case TYPE_SPECIFIC_SELF_TYPE:
@@ -5633,8 +5664,9 @@ copy_type_recursive (struct type *type, htab_t copied_types)
case TYPE_SPECIFIC_FUNC:
INIT_FUNC_SPECIFIC (new_type);
TYPE_CALLING_CONVENTION (new_type) = TYPE_CALLING_CONVENTION (type);
- TYPE_NO_RETURN (new_type) = TYPE_NO_RETURN (type);
+ TYPE_FUNC_FLAGS (new_type) = TYPE_FUNC_FLAGS (type);
TYPE_TAIL_CALL_LIST (new_type) = NULL;
+ TYPE_TRAMPOLINE_TARGET (new_type) = TYPE_TRAMPOLINE_TARGET (type);
break;
case TYPE_SPECIFIC_FLOATFORMAT:
TYPE_FLOATFORMAT (new_type) = TYPE_FLOATFORMAT (type);
diff --git a/gdb/gdbtypes.h b/gdb/gdbtypes.h
index aedaf53cd5d..e43cd4e4d3b 100644
--- a/gdb/gdbtypes.h
+++ b/gdb/gdbtypes.h
@@ -106,6 +106,21 @@ enum type_instance_flag_value : unsigned
DEF_ENUM_FLAGS_TYPE (enum type_instance_flag_value, type_instance_flags);
+/* * Define flags for function types. */
+enum func_type_flag_value : unsigned
+{
+ /* * Flag indicates, whether this function normally returns to its
+ caller. It is set from the DW_AT_noreturn attribute if set on
+ the DW_TAG_subprogram. */
+ FUNC_TYPE_NO_RETURN = (1 << 0),
+
+ /* * Flag is used for functions marked with DW_AT_trampoline. These
+ are compiler generated wrappers that should be hidden from the user. */
+ FUNC_TYPE_TRAMPOLINE = (1 << 1)
+};
+
+DEF_ENUM_FLAGS_TYPE (enum func_type_flag_value, func_type_flags);
+
/* * Not textual. By default, GDB treats all single byte integers as
characters (or elements of strings) unless this flag is set. */
@@ -1730,11 +1745,9 @@ struct func_type
ENUM_BITFIELD (dwarf_calling_convention) calling_convention : 8;
- /* * Whether this function normally returns to its caller. It is
- set from the DW_AT_noreturn attribute if set on the
- DW_TAG_subprogram. */
+ /* * For storing function types defined in eunm func_type_flag_value. */
- unsigned int is_noreturn : 1;
+ func_type_flags flags;
/* * Only those DW_TAG_call_site's in this function that have
DW_AT_call_tail_call set are linked in this list. Function
@@ -1749,6 +1762,87 @@ struct func_type
contains the method. */
struct type *self_type;
+
+ struct trampoline_target *self_trampoline_target;
+ };
+
+/* The kind of location held by this call site target. */
+enum trampoline_target_kind
+ {
+ /* An address. */
+ TRAMPOLINE_TARGET_ADDR,
+ /* A (mangled) name. */
+ TRAMPOLINE_TARGET_NAME,
+ /* A flag (target is unknown). */
+ TRAMPOLINE_TARGET_FLAG,
+ };
+
+/* Target information for trampoline functions. */
+
+struct trampoline_target
+ {
+ /* Returns what kind of target the trampoline points to. */
+ trampoline_target_kind target_kind () const
+ {
+ return m_target_kind;
+ }
+
+ /* Set the address for trampoline target. */
+ void set_target_addr (unrelocated_addr unreloc_addr)
+ {
+ m_target_kind = TRAMPOLINE_TARGET_ADDR;
+ m_trampoline_target.unreloc_addr = unreloc_addr;
+ }
+
+ /* Return the address of the trampoline target. */
+ unrelocated_addr target_addr () const
+ {
+ gdb_assert (m_target_kind == TRAMPOLINE_TARGET_ADDR);
+ return m_trampoline_target.unreloc_addr;
+ }
+
+ /* Set the name of the trampoline target. */
+ void set_target_name (const char *name)
+ {
+ m_target_kind = TRAMPOLINE_TARGET_NAME;
+ m_trampoline_target.name = name;
+ }
+
+ /* Return the name of the trampoline target. */
+ const char *target_name () const
+ {
+ gdb_assert (m_target_kind == TRAMPOLINE_TARGET_NAME);
+ return m_trampoline_target.name;
+ }
+
+ /* Set the flag for trampoline target subroutine. */
+ void set_target_flag (bool flag)
+ {
+ m_target_kind = TRAMPOLINE_TARGET_FLAG;
+ m_trampoline_target.flag = flag;
+ }
+
+ /* Return the flag for trampoline target. */
+ bool target_flag () const
+ {
+ gdb_assert (m_target_kind == TRAMPOLINE_TARGET_FLAG);
+ return m_trampoline_target.flag;
+ }
+
+ private:
+
+ union
+ {
+ /* Address. */
+ unrelocated_addr unreloc_addr;
+ /* Mangled name. */
+ const char *name;
+ /* Flag. */
+ bool flag;
+ } m_trampoline_target;
+
+ /* * Discriminant for union m_trampoline_target. */
+ ENUM_BITFIELD (trampoline_target_kind) m_target_kind : 2;
};
/* The type-specific info for TYPE_CODE_FIXED_POINT types. */
@@ -1891,7 +1985,16 @@ extern void set_type_vptr_basetype (struct type *, struct type *);
#define TYPE_GNAT_SPECIFIC(thistype) TYPE_MAIN_TYPE(thistype)->type_specific.gnat_stuff
#define TYPE_DESCRIPTIVE_TYPE(thistype) TYPE_GNAT_SPECIFIC(thistype)->descriptive_type
#define TYPE_CALLING_CONVENTION(thistype) TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->calling_convention
-#define TYPE_NO_RETURN(thistype) TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->is_noreturn
+#define TYPE_FUNC_FLAGS(thistype) \
+ TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->flags
+#define TYPE_NO_RETURN(thistype) \
+ (TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->flags \
+ & FUNC_TYPE_NO_RETURN)
+#define TYPE_IS_TRAMPOLINE(thistype) \
+ (TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->flags \
+ & FUNC_TYPE_TRAMPOLINE)
+#define TYPE_TRAMPOLINE_TARGET(thistype) \
+ TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->self_trampoline_target
#define TYPE_TAIL_CALL_LIST(thistype) TYPE_MAIN_TYPE(thistype)->type_specific.func_stuff->tail_call_list
#define TYPE_BASECLASS(thistype,index) ((thistype)->field (index).type ())
#define TYPE_N_BASECLASSES(thistype) TYPE_CPLUS_SPECIFIC(thistype)->n_baseclasses
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v5 2/4] gdb/symtab: add lookup for trampoline functions
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 1/4] gdb, dwarf: add support for DW_AT_trampoline in DWARF reader Abdul Basit Ijaz
@ 2023-08-16 21:53 ` Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 3/4] gdb/infrun: handle stepping through functions with DW_AT_trampoline Abdul Basit Ijaz
` (3 subsequent siblings)
5 siblings, 0 replies; 13+ messages in thread
From: Abdul Basit Ijaz @ 2023-08-16 21:53 UTC (permalink / raw)
To: gdb-patches; +Cc: abdul.b.ijaz, JiniSusan.George, tom, eliz, blarsen
In order to query information about the DW_AT_trampoline tag for
subroutines and inlined subroutines, two function were added to symtab.
First, a routine for querying whether the given pc belongs to a block
that is associated with a function (maybe inlined) marked
DW_AT_trampoline.
Second, a routine for querying a trampoline function's target.
Subroutines and inlined subroutines marked with DW_AT_trampoline usually
contain information about the target subroutine they are 'wrapping'/
passing control to.
2023-08-16 Nils-Christian Kempke <nils-christian.kempke@intel.com>
---
gdb/symtab.c | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++++
gdb/symtab.h | 14 +++++++++++
2 files changed, 82 insertions(+)
diff --git a/gdb/symtab.c b/gdb/symtab.c
index 0117a2a59d7..855f9b41ec7 100644
--- a/gdb/symtab.c
+++ b/gdb/symtab.c
@@ -73,6 +73,7 @@
#include "gdbsupport/gdb_string_view.h"
#include "gdbsupport/pathstuff.h"
#include "gdbsupport/common-utils.h"
+#include "gdbsupport/symbol.h"
/* Forward declarations for local functions. */
@@ -4051,6 +4052,73 @@ find_function_alias_target (bound_minimal_symbol msymbol)
return NULL;
}
+/* See symtab.h. */
+
+bool
+in_trampoline_function (CORE_ADDR pc)
+{
+ /* Find the innermost function containing pc. This might be an inlined
+ function. */
+ symbol *sym = find_pc_sect_containing_function (pc,
+ find_pc_mapped_section (pc));
+ return sym != nullptr && TYPE_IS_TRAMPOLINE (sym->type ());
+}
+
+/* See symtab.h. */
+
+CORE_ADDR
+find_function_trampoline_target (CORE_ADDR pc)
+{
+ /* Find the innermost function containing pc. This might be an inlined
+ function. */
+ symbol *sym = find_pc_sect_containing_function (pc,
+ find_pc_mapped_section (pc));
+ CORE_ADDR target_address = 0;
+
+ if (sym != nullptr && TYPE_IS_TRAMPOLINE (sym->type ()))
+ {
+ trampoline_target *trampoline = TYPE_TRAMPOLINE_TARGET (sym->type ());
+
+ /* DW_AT_trampoline can be given as an address, name, or flag here (die
+ references have been resolved as names at this point. In the case
+ where DW_AT_trampoline contains a flag we do not know the target
+ address and return 0. */
+ if (trampoline->target_kind () == TRAMPOLINE_TARGET_NAME)
+ {
+ /* Handle both the mangled and demangled PHYSNAME. */
+ const char *physname = trampoline->target_name ();
+
+ /* First, check whether there exists a symbol matching the
+ physname. If we cannot find one also check for minimal
+ symbols. */
+ const block *blk = block_for_pc (pc);
+ struct block_symbol bs = lookup_symbol (physname, blk, VAR_DOMAIN, 0);
+ if (bs.symbol != nullptr)
+ {
+ const struct block *block = bs.symbol->value_block ();
+ gdb_assert (block != nullptr);
+ target_address = block->start ();
+ }
+ else
+ {
+ if (find_minimal_symbol_address (physname, &target_address,
+ nullptr) != 0)
+ target_address = 0;
+ }
+ }
+ else if (trampoline->target_kind () == TRAMPOLINE_TARGET_ADDR)
+ {
+ /* If the function symbol containing this trampoline target has
+ been relocated we assume the target_address also needs relocation.
+ If it has not been relocated the offset should be zero. */
+ target_address = \
+ ( (CORE_ADDR) trampoline->target_addr ()
+ + sym->objfile ()->section_offsets[sym->section_index ()]);
+ }
+ }
+
+ return target_address;
+}
\f
/* If P is of the form "operator[ \t]+..." where `...' is
some legitimate operator text, return a pointer to the
diff --git a/gdb/symtab.h b/gdb/symtab.h
index ee4729b14cd..fc2d82e6567 100644
--- a/gdb/symtab.h
+++ b/gdb/symtab.h
@@ -2312,6 +2312,20 @@ extern const struct gnu_ifunc_fns *gnu_ifunc_fns_p;
extern CORE_ADDR find_solib_trampoline_target (frame_info_ptr, CORE_ADDR);
+/* Return whether or not the current pc is within a block that belongs to a
+ function that is marked as a trampoline by the compiler. */
+
+extern bool in_trampoline_function (CORE_ADDR pc);
+
+/* Find the target of a trampoline function marked via the DW_AT_trampoline
+ attribute and return its address. Returns 0 if the pc is not contained
+ in a trampoline function (inlined or not). If DW_AT_trampoline
+ is given as a flag, the target is unknown and the function will still return
+ 0. One has to additionally query in_trampoline_function to cover this
+ case. */
+
+extern CORE_ADDR find_function_trampoline_target (CORE_ADDR pc);
+
struct symtab_and_line
{
/* The program space of this sal. */
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v5 3/4] gdb/infrun: handle stepping through functions with DW_AT_trampoline
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 1/4] gdb, dwarf: add support for DW_AT_trampoline in DWARF reader Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 2/4] gdb/symtab: add lookup for trampoline functions Abdul Basit Ijaz
@ 2023-08-16 21:53 ` Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 4/4] gdb: Skip trampoline frames in the stack for printing or finish command Abdul Basit Ijaz
` (2 subsequent siblings)
5 siblings, 0 replies; 13+ messages in thread
From: Abdul Basit Ijaz @ 2023-08-16 21:53 UTC (permalink / raw)
To: gdb-patches; +Cc: abdul.b.ijaz, JiniSusan.George, tom, eliz, blarsen
This patch makes infrun continue stepping into and through trampoline
functions marked via the DW_AT_trampoline in DWARF. The attribute can
be emitted by the compiler for certain subroutines/inlined subroutines
that are compiler generated and should be hidden from a user.
Mainly, infrun is modified in 3 ways.
First, GDB will now attempt to step through trampoline functions.
Whenever we issued a step command that would make GDB step into a
function that is marked trampoline, GDB will try to step directly towards
the trampoline's 'target' instead and, e.g., not stop at the first
instruction of the trampoline.
The target can be specified by the compiler by the value of
DW_AT_trampoline if its form is either an address, a name, or a DIE
reference. DW_AT_trampoline is also allowed to be specified as a flag
(containing true or false), in which case the target is assumed to be
unknown. If GDB successfully finds a target, so if the value of
DW_AT_trampoline was not a flag and could be resolved successfully, GDB
steps directly towards the target and through the trampoline, hiding
the trampoline from the user. If GDB cannot, however deduce a target,
most likely because the DW_AT_trampoline was given as a flag or because
of broken debug info, it will instead continue inside execution in the
trampoline function until it reaches an instruction that is not
associated with a trampoline function, which is usually the target
function. It will then stop and give control back to the user.
It should be noted, that there might be the cases, where trampolines
call functions other than the target before the actual target call. If,
in such a situation, GDB fails to resolve the target, it would resume
execution until stepping into this other function call, and hand back
control to the user, without actually having reached the target. A
second step would have to be issued by the user to arrive a the target
(by resuming in the trampoline and then until leaving it a second time).
As this is a rather pathological case and no real instance of this is
known, I think the current behavior here is good enough and seems to be
the best GDB can do in such a situation.
Secondly, as trampoline functions normally do not have any real source
code correlation, it is likely that they mostly appear without line
info.
Normally, GDB would skip completely over a function call to a function
that has no source line information, so we would never get to the
aforementioned stepping through a trampoline and target resolution. To
remedy this, for debug info trampolines, GDB now attempts to step through
them regardless of them having source line information or not. So
issuing a step at a function call wrapped by a trampoline without source
line information will no longer skip the whole function call, but now
step through the trampoline and attempt to resolve the trampoline target
as described above (so usually, a single step at the call site will step
through the trampoline and towards the target, even if the trampoline had
not source line info).
Last, in all other cases when GDB is about to stop at a location that is
included in a trampoline region (e.g. after a step from the target back
into the trampoline) GDB will instead continue until the trampoline
region is left again and only then give control back to the user. This
change serves the purpose of allowing stepping back from a target call
through the trampoline without the user noticing the artificial function
call inbetween call site and target.
Together, these changes attempt to hide the trampoline function from the
user while stepping. Additionally, the skip-trampoline-functions option
has been introduced in infrun. It is set by default, and, when turned
off, GDB will return to its 'normal' stepping behavior and ignore any
possible DW_AT_trampoline.
As currently only ifx emits the DW_AT_trampoline tag, a test has been
added to gdb.dwarf2 that artificially creates a set of trampoline
functions.
2023-08-16 Nils-Christian Kempke <nils-christian.kempke@intel.com>
---
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 36 +++
gdb/infrun.c | 85 +++++-
gdb/infrun.h | 4 +
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
6 files changed, 461 insertions(+), 4 deletions(-)
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
diff --git a/gdb/NEWS b/gdb/NEWS
index ef2b3b3a4f5..1756a2769ff 100644
--- a/gdb/NEWS
+++ b/gdb/NEWS
@@ -598,6 +598,21 @@ platform.
* New commands
+set skip-trampoline-functions on|off
+show skip-trampoline-functions
+ This setting is 'on' by default. When 'on' it controls whether GDB's
+ stepping behavior will recognize function calls that have been marked as
+ trampolines in the debug info. It improves stepping behavior in that it
+ steps through trampoline code and hides it from the user. GDB can now step
+ through trampolines that are correctly marked as such in the compiler's
+ debug info. If the target of a trampoline is unknown, GDB will continue
+ until the trampoline section is left again and only then hand control back
+ to the user. GDB does this even if the trampoline has no associated line
+ info. If this is turned off, GDB will step into trampolines if there is
+ line table information for them or step over the trampoline calls if there
+ is no line table information. Currently, only DWARF trampolines are
+ supported.
+
maint set backtrace-on-fatal-signal on|off
maint show backtrace-on-fatal-signal
This setting is 'on' by default. When 'on' GDB will print a limited
diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
index 814cc6d714a..abebad04632 100644
--- a/gdb/doc/gdb.texinfo
+++ b/gdb/doc/gdb.texinfo
@@ -6391,6 +6391,42 @@ debug information. This is the default.
Show whether @value{GDBN} will stop in or step over functions without
source line debug information.
+@kindex set skip-trampoline-functions
+@cindex trampoline functions
+@cindex stepping through trampoline functions
+@item set skip-trampoline-functions
+@itemx set skip-trampoline-functions on
+When calling a function in any language, some compilers might generate
+so-called @dfn{trampoline functions}, which wrap the actual function call
+(the target of the trampoline). The compiler might mark such a trampoline in
+its debug information. Often, such trampolines do not have any source line
+information associated with them which will lead the @code{step} command to
+behave like a @code{next} and skip the function call completely.
+
+The default is @code{set skip-trampoline-functions on} and it will cause the
+@code{step} command to treat these trampolines differently. When issuing a
+@code{step} at the call site of a trampoline function if
+@code{skip-trampoline-functions} is set @value{GDBN} will attempt to determine
+the target of the trampoline and then step through the trampoline stopping at
+the target. If the target could not be found or was not given in the debug
+info, @value{GDBN} will simply continue execution until it leaves the
+trampoline code again, even if the trampoline has no line info associated with
+it. When returning from a target function call and stepping back into the
+trampoline, @value{GDBN} will again step through the trampoline towards the
+call site. Additionally, even if stopped in a trampoline function with source
+line information, issuing a @code{step} will prompt @value{GDBN} to resume
+execution until leaving the trampoline region again. The @code{stepi} command
+is not affected by the setting which is enabled by default. Currently, only
+DWARF trampolines marked via DW_AT_trampoline are supported by this.
+
+@item set skip-trampoline-functions off
+Causes the @code{step} command to completely ignore any trampoline information
+a compiler might have emitted in its debug info. Trampolines will be treated
+like any other function when stepping.
+
+@item show skip-trampoline-functions
+Show whether @value{GDBN} tries to skip trampolines or not.
+
@kindex finish
@kindex fin @r{(@code{finish})}
@item finish
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 8286026e6c6..b2597681e00 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -184,6 +184,12 @@ get_previous_thread ()
static bool detach_fork = true;
+/* If set (default) GDB will step through functions/inlined subroutines marked
+ DW_AT_trampoline by the compiler. If false, GDB will ignore the
+ attribute. */
+
+static bool skip_trampoline_functions = true;
+
bool debug_infrun = false;
static void
show_debug_infrun (struct ui_file *file, int from_tty,
@@ -7342,7 +7348,6 @@ process_event_stop_test (struct execution_control_state *ecs)
!= find_pc_function (ecs->event_thread->stop_pc ())))))
{
CORE_ADDR stop_pc = ecs->event_thread->stop_pc ();
- CORE_ADDR real_stop_pc;
infrun_debug_printf ("stepped into subroutine");
@@ -7414,8 +7419,45 @@ process_event_stop_test (struct execution_control_state *ecs)
calling routine and the real function), locate the real
function. That's what tells us (a) whether we want to step
into it at all, and (b) what prologue we want to run to the
- end of, if we do step into it. */
- real_stop_pc = skip_language_trampoline (frame, stop_pc);
+ end of, if we do step into it. For functions marked as
+ trampoline functions we try to find their target and step
+ towards it (if skip_trampoline_functions is not set to false by the
+ user). If no target can be determined we just step into the
+ trampoline and hand control back to the user. */
+ CORE_ADDR real_stop_pc = 0;
+
+ if (skip_trampoline_functions && in_trampoline_function (stop_pc))
+ {
+ if (execution_direction == EXEC_REVERSE)
+ {
+ keep_going (ecs);
+ return;
+ }
+ real_stop_pc = find_function_trampoline_target (stop_pc);
+
+ for (int i = 0;
+ (i < MAX_TRAMPOLINE_CHAIN_SIZE
+ && in_trampoline_function (real_stop_pc)); ++i)
+ {
+ real_stop_pc = find_function_trampoline_target (real_stop_pc);
+ /* Exit if find_function_trampoline_target failed to find the
+ trampoline target. Do not try to resolve the trampolines
+ in this case. */
+ if (real_stop_pc == 0)
+ break;
+ }
+
+ /* If we failed to find a target we will just single step in the
+ hope of leaving the trampoline again soon. */
+ if (real_stop_pc == 0)
+ {
+ keep_going (ecs);
+ return;
+ }
+ }
+
+ if (real_stop_pc == 0)
+ real_stop_pc = skip_language_trampoline (frame, stop_pc);
if (real_stop_pc == 0)
real_stop_pc = gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc);
if (real_stop_pc != 0)
@@ -7436,6 +7478,8 @@ process_event_stop_test (struct execution_control_state *ecs)
/* If we have line number information for the function we are
thinking of stepping into and the function isn't on the skip
list, step into it.
+ If we are about to step into a function marked trampoline with no
+ line number information, we still want to enter it here.
If there are several symtabs at that PC (e.g. with include
files), just want to know whether *any* of them have line
@@ -7444,7 +7488,9 @@ process_event_stop_test (struct execution_control_state *ecs)
struct symtab_and_line tmp_sal;
tmp_sal = find_pc_line (ecs->stop_func_start, 0);
- if (tmp_sal.line != 0
+ if ((tmp_sal.line != 0
+ || (skip_trampoline_functions
+ && in_trampoline_function (ecs->stop_func_start)))
&& !function_name_is_marked_for_skip (ecs->stop_func_name,
tmp_sal)
&& !inline_frame_is_marked_for_skip (true, ecs->event_thread))
@@ -7597,6 +7643,19 @@ process_event_stop_test (struct execution_control_state *ecs)
return;
}
+ /* If we ended up in a function trampoline without stepping into a new
+ function we are either in some inlined trampoline or returning through a
+ trampoline function. In either case we continue to single step until we
+ are out of the trampoline code again. This check has to be done before
+ stop_pc_sal.line == 0 below, as trampolines usually don't have source
+ line information associated with them. */
+ if (skip_trampoline_functions && in_trampoline_function (stop_pc_sal.pc))
+ {
+ infrun_debug_printf ("stepped into trampoline code");
+ keep_going (ecs);
+ return;
+ }
+
if (stop_pc_sal.line == 0)
{
/* We have no line number information. That means to stop
@@ -9808,6 +9867,14 @@ show_exec_direction_func (struct ui_file *out, int from_tty,
}
}
+static void
+show_skip_trampoline_functions (ui_file *file, int from_tty,
+ cmd_list_element *c,
+ const char *value)
+{
+ gdb_printf (file, _("Skipping trampoline functions is %s.\n"), value);
+}
+
static void
show_schedule_multiple (struct ui_file *file, int from_tty,
struct cmd_list_element *c, const char *value)
@@ -10153,6 +10220,16 @@ Options are 'forward' or 'reverse'."),
set_exec_direction_func, show_exec_direction_func,
&setlist, &showlist);
+ add_setshow_boolean_cmd ("skip-trampoline-functions", class_run,
+ &skip_trampoline_functions, _("\
+Set whether gdb attempts to hide trampolines marked in the debug info."), _("\
+Show whether gdb attempts to hide trampolines marked in the debug info."), _("\
+If on, while stepping gdb will skip through functions and inlined functions\n\
+marked as trampolines by the compiler. If off, gdb will ignore such function\n\
+trampolines."),
+ nullptr, show_skip_trampoline_functions, &setlist,
+ &showlist);
+
/* Set/show detach-on-fork: user-settable mode. */
add_setshow_boolean_cmd ("detach-on-fork", class_run, &detach_fork, _("\
diff --git a/gdb/infrun.h b/gdb/infrun.h
index a343d27f72d..d4a6b7892b5 100644
--- a/gdb/infrun.h
+++ b/gdb/infrun.h
@@ -76,6 +76,10 @@ infrun_debug_show_threads (const char *title, ThreadRange threads)
}
+/* Maximum size of trampoline chain to process while resolving
+ trampolines. */
+#define MAX_TRAMPOLINE_CHAIN_SIZE 10
+
/* Nonzero if we want to give control to the user when we're notified
of shared library events by the dynamic linker. */
extern int stop_on_solib_events;
diff --git a/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c b/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
new file mode 100644
index 00000000000..1f0eb2e0efa
--- /dev/null
+++ b/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
@@ -0,0 +1,80 @@
+/* Copyright 2023 Free Software Foundation, Inc.
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
+
+/* This test relies on inlined_trampoline being inlined into main and the other
+ functions not. All functions except target will be marked via
+ DW_AT_trampoline in the debug info and we'll check whether one can step
+ through the trampolines towards target. */
+volatile int global_var;
+
+int __attribute__ ((noinline))
+target () /* target decl line */
+{ /* target prologue */
+ asm ("target_label: .globl target_label");
+ ++global_var; /* target add */
+ asm ("target_label2: .globl target_label2");
+ return 9 + 10; /* target return */
+} /* target end */
+
+int __attribute__ ((noinline))
+trampoline ()
+{ /* trampoline prologue */
+ asm ("trampoline_label: .globl trampoline_label");
+ ++global_var;
+ return target (); /* trampoline target call */
+} /* trampoline end */
+
+static inline int __attribute__ ((always_inline))
+inlined_trampoline ()
+{ /* inlined_trampoline prologue */
+ asm ("inlined_trampoline_label: .globl inlined_trampoline_label");
+ ++global_var; /* inlined_trampoline add */
+ asm ("inlined_trampoline_label2: .globl inlined_trampoline_label2");
+ return target (); /* inlined_trampoline target call */
+} /* inlined_trampoline end */
+
+int __attribute__ ((noinline))
+chained_trampoline ()
+{ /* chained_trampoline prologue */
+ asm ("chained_trampoline_label: .globl chained_trampoline_label");
+ ++global_var;
+ return trampoline (); /* chained_trampoline trampoline call */
+} /* chained_trampoline end */
+
+int __attribute__ ((noinline))
+doubly_chained_trampoline ()
+{ /* doubly_chained_trampoline prologue */
+ asm ("doubly_chained_trampoline_label: .globl doubly_chained_trampoline_label");
+ ++global_var;
+ return chained_trampoline (); /* doubly_chained_trampoline chained_trampoline call */
+} /* doubly_chained_trampoline end */
+
+int
+main () /* main decl line */
+{ /* main prologue */
+ int ans;
+ asm ("main_label: .globl main_label");
+ global_var = 0; /* main set global_var */
+ asm ("main_label2: .globl main_label2");
+ ans = inlined_trampoline (); /* main call inlined_trampoline */
+ asm ("main_label3: .globl main_label3");
+ ans = trampoline (); /* main call trampoline */
+ asm ("main_label4: .globl main_label4");
+ ans = chained_trampoline (); /* main call chained_trampoline */
+ asm ("main_label5: .globl main_label5");
+ ans = doubly_chained_trampoline (); /* main call doubly_chained_trampoline */
+ asm ("main_label6: .globl main_label6");
+ return ans; /* main call return */
+} /* main end */
diff --git a/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp b/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
new file mode 100644
index 00000000000..67eec2b5540
--- /dev/null
+++ b/gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
@@ -0,0 +1,245 @@
+# Copyright 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+
+# This test checks GDB's handling of functions/inlined functions marked
+# DW_AT_trampoline by the compiler. A function marked as trampoline should
+# generally be hidden from the user. We check whether we can step through
+# trampolines. Every trampoline is defined using a different type for its
+# target: a string, an address, a DIE reference or a flag.
+# Setting skip-trampoline-functions to false inside GDB should make it return
+# to its 'normal' behavior, ignore the DW_AT_trampoline, and skip all of the
+# non-inlined trampoline calls (as their DIEs don't have any source
+# information).
+
+load_lib dwarf.exp
+
+# This test can only be run on targets which support DWARF-2 and use gas.
+if {![dwarf2_support]} {
+ return 0
+}
+
+standard_testfile .c .S
+
+set asm_file [standard_output_file $srcfile2]
+Dwarf::assemble $asm_file {
+ global srcdir subdir srcfile
+ declare_labels lines_label trampoline_label
+
+ get_func_info target
+ get_func_info trampoline
+ get_func_info chained_trampoline
+ get_func_info doubly_chained_trampoline
+ get_func_info main
+
+ set target_decl_line [gdb_get_line_number "target decl line"]
+ set main_decl_line [gdb_get_line_number "main decl line"]
+ set main_call_inlined_trampoline_line [gdb_get_line_number "main call inlined_trampoline"]
+
+ cu {} {
+ compile_unit {
+ {language @DW_LANG_C}
+ {name dw2-function-trampolines.c}
+ {low_pc 0 addr}
+ {stmt_list ${lines_label} DW_FORM_sec_offset}
+ } {
+ subprogram {
+ {name target}
+ {low_pc $target_start addr}
+ {high_pc "$target_start + $target_len" addr}
+ {decl_file 1 data1}
+ {decl_line $target_decl_line data1}
+ }
+ # The 'trampoline' subprogram declares its target by name.
+ trampoline_label: subprogram {
+ {name trampoline}
+ {low_pc $trampoline_start addr}
+ {high_pc "$trampoline_start + $trampoline_len" addr}
+ {trampoline target string}
+ }
+ # The 'chained_trampoline' subprogram declares its target as die
+ # reference.
+ subprogram {
+ {name chained_trampoline}
+ {low_pc $chained_trampoline_start addr}
+ {high_pc "$chained_trampoline_start + $chained_trampoline_len" addr}
+ {trampoline %$trampoline_label}
+ }
+ # The 'doubly_chained_trampoline' subprogram declares no target.
+ # Its DW_AT_trampoline is a flag set to true.
+ subprogram {
+ {name doubly_chained_trampoline}
+ {low_pc $doubly_chained_trampoline_start addr}
+ {high_pc "$doubly_chained_trampoline_start + $doubly_chained_trampoline_len" addr}
+ {trampoline 1 flag}
+ }
+ subprogram {
+ {external 1 flag}
+ {name main}
+ {main_subprogram 1 flag}
+ {low_pc $main_start addr}
+ {high_pc "$main_start + $main_len" addr}
+ {decl_file 1 data1}
+ {decl_line $main_decl_line data1}
+ } {
+ # The 'inlined_trampoline' subroutine declares its target as
+ # an address.
+ inlined_subroutine {
+ {name inlined_trampoline}
+ {low_pc main_label2 addr}
+ {high_pc main_label3 addr}
+ {trampoline $target_start addr}
+ {call_file 1 data1}
+ {call_line $main_call_inlined_trampoline_line data1}
+ }
+ }
+ }
+ }
+
+ lines {version 2} lines_label {
+ include_dir "${srcdir}/${subdir}"
+ file_name "$srcfile" 1
+
+ program {
+ DW_LNE_set_address $main_start
+ line [gdb_get_line_number "main set global_var"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label
+ line [gdb_get_line_number "main set global_var"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label2
+ line [gdb_get_line_number "main call inlined_trampoline"]
+ DW_LNS_copy
+ DW_LNE_set_address inlined_trampoline_label
+ line [gdb_get_line_number "inlined_trampoline add"]
+ DW_LNS_copy
+ DW_LNE_set_address inlined_trampoline_label2
+ line [gdb_get_line_number "inlined_trampoline target call"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label3
+ line [gdb_get_line_number "main call trampoline"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label4
+ line [gdb_get_line_number "main call chained_trampoline"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label5
+ line [gdb_get_line_number "main call doubly_chained_trampoline"]
+ DW_LNS_copy
+ DW_LNE_set_address main_label6
+ line [gdb_get_line_number "main call return"]
+ DW_LNS_copy
+ DW_LNE_set_address $main_end
+ DW_LNE_end_sequence
+
+ DW_LNE_set_address $target_start
+ line [gdb_get_line_number "target prologue"]
+ DW_LNS_negate_stmt
+ DW_LNS_copy
+ DW_LNE_set_address target_label
+ line [gdb_get_line_number "target add"]
+ DW_LNS_negate_stmt
+ DW_LNS_copy
+ DW_LNE_set_address target_label2
+ line [gdb_get_line_number "target return"]
+ DW_LNS_copy
+ DW_LNE_set_address $target_end
+ DW_LNE_end_sequence
+ }
+ }
+}
+
+if {[prepare_for_testing "failed to prepare" ${testfile} \
+ [list $srcfile $asm_file] {nodebug additional_flags=-O0}]} {
+ return -1
+}
+
+set target_first_line_pattern ".*target add.*"
+set target_second_line_pattern ".*target return.*"
+
+if ![runto_main] {
+ return -1
+}
+
+gdb_test "show skip-trampoline-functions" \
+ "Skipping trampoline functions is on\." \
+ "check skip-trampoline-functions is enabled"
+
+with_test_prefix "with trampoline handling" {
+ foreach {trampoline return_line} { "inlined_trampoline" "trampoline" \
+ "trampoline" "chained_trampoline" \
+ "chained_trampoline" "doubly_chained_trampoline" } {
+
+ gdb_test "s" "$target_first_line_pattern" "step through $trampoline"
+ gdb_test "s" "$target_second_line_pattern" \
+ "step target second line from $trampoline"
+ gdb_test "s" ".*main call $return_line.*" \
+ "step back through $trampoline"
+ }
+
+ # The doubly_chained_trampoline has only been marked as trampoline but no
+ # target was given. In this case GDB steps into the trampoline and then
+ # continues until the trampoline section is left again.
+
+ # When compiled with gcc 7.5 (and possibly others) on a 32 bit system, the
+ # trampoline function contains a call to __x86.get_pc_thunk.ax before the
+ # actual target call. So, we end up in __x86.get_pc_thunk.ax. Issuing a
+ # second step command will return from the function call back into the
+ # trampoline and go on inside the trampoline towards the actual target call.
+ # On other targets we step directly towards the target call.
+ gdb_test_multiple "s" "step through double_chained_trampoline" {
+ -re -wrap "$target_first_line_pattern" {
+ pass $gdb_test_name
+ }
+ -re -wrap ".*__x86.get_pc_thunk.ax.*" {
+ gdb_test "s" "$target_first_line_pattern" \
+ "step through double_chained_trampoline 2nd try"
+ }
+ }
+ gdb_test "s" "$target_second_line_pattern" \
+ "step target second line fromdoubly_chained_trampoline"
+ gdb_test "s" ".*main call return.*" \
+ "step back through doubly_chained_trampoline"
+}
+
+clean_restart ${testfile}
+
+if ![runto_main] {
+ return -1
+}
+
+gdb_test_no_output "set skip-trampoline-functions off" \
+ "disable trampoline handling"
+gdb_test "show skip-trampoline-functions" \
+ "Skipping trampoline functions is off." \
+ "check skip-trampoline-functions is disabled"
+
+with_test_prefix "without trampoline handling" {
+ gdb_test "s" ".*main call inlined_trampoline.*"
+ gdb_test "s" ".*inlined_trampoline add.*" \
+ "step into inlined_trampoline with skip-trampoline off"
+ gdb_test "s" ".*inlined_trampoline target call.*" \
+ "step in inlined_trampoline with skip-trampoline off"
+ gdb_test "s" "$target_first_line_pattern" \
+ "step into target with skip-trampoline off"
+ gdb_test "s" "$target_second_line_pattern" \
+ "step second line in target with skip-trampoline off"
+ gdb_test "s" ".*main call trampoline.*" \
+ "step brack from target with skip-trampoline off"
+ gdb_test "s" ".*main call chained_trampoline.*" \
+ "skip trampoline call with no line info"
+ gdb_test "s" ".*main call doubly_chained_trampoline.*" \
+ "skip chained_trampoline call with no line info"
+ gdb_test "s" ".*main call return.*" \
+ "skip doubly_chained_trampoline call with no line info"
+}
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v5 4/4] gdb: Skip trampoline frames in the stack for printing or finish command.
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
` (2 preceding siblings ...)
2023-08-16 21:53 ` [PATCH v5 3/4] gdb/infrun: handle stepping through functions with DW_AT_trampoline Abdul Basit Ijaz
@ 2023-08-16 21:53 ` Abdul Basit Ijaz
2023-08-17 5:26 ` [PATCH v5 0/4] GDB support for DW_AT_trampoline Eli Zaretskii
2023-09-03 19:14 ` [PING][PATCH " Ijaz, Abdul B
5 siblings, 0 replies; 13+ messages in thread
From: Abdul Basit Ijaz @ 2023-08-16 21:53 UTC (permalink / raw)
To: gdb-patches; +Cc: abdul.b.ijaz, JiniSusan.George, tom, eliz, blarsen
Before the change, GDB prints the frames indicated by the compiler with DIE
"DW_AT_trampoline" in the backtrace and finish command, but for better user
experience, all such frames can be hidden from the user. So, after this
change, now such frames are not printed any more in the backtrace command
and also the 'finish' command skips the trampoline calls.
So far, this DIE is added to DWARF only by the IFX compiler, so
gdb.fortran/mixed-lang-stack test used to fail for this compiler because of
these extra trampoline frames in the backtrace. After the commit, those
trampoline frames are filtered so test is updated accordingly to handle the
frame level of the filtered frames.
Backtrace output in this test before the change:
bt -frame-arguments all
...
...
(gdb) FAIL: gdb.fortran/mixed-lang-stack.exp: lang=auto: bt -frame-arguments all
Now, after the change:
bt -frame-arguments all
...
...
(gdb) PASS: gdb.fortran/mixed-lang-stack.exp: lang=auto: bt -frame-arguments all
Below is the example where IFX emits DW_at_trampoline for functions first
and second trampoline calls as following:
F90 reproducer info:
function second (x, y) result(z)
integer, intent(in) :: x, y
integer :: z
z = x * y ! breakpt-backtrace
end function second
function first (num1, num2) result(total)
integer, intent(in) :: num1, num2
integer :: total
total = second (num1 + 4, num2 * 3) ! first-breakpt
total = total + 30
end function first
Related Dwarf:
0x0000013f: DW_TAG_subprogram
DW_AT_low_pc (0x0000000000404350)
DW_AT_high_pc (0x000000000040435f)
DW_AT_frame_base (DW_OP_reg6 RBP)
DW_AT_linkage_name ("second_.t74p.t75p")
DW_AT_name ("second_.t74p.t75p")
DW_AT_trampoline ("second_")
0x0000015a: DW_TAG_subprogram
DW_AT_low_pc (0x00000000004044a0)
DW_AT_high_pc (0x00000000004044af)
DW_AT_frame_base (DW_OP_reg6 RBP)
DW_AT_linkage_name ("first_.t104p.t105p")
DW_AT_name ("first_.t104p.t105p")
DW_AT_trampoline ("first_")
Before this change at line with comment 'breakpt-backtrace':
(gdb) backtrace 3
(gdb) finish
Run till exit from #0 second (x=20, y=9) at test.f90:4
second_.t74p.t75p () at test.f90:10
10 total = second (num1 + 4, num2 * 3) ! first-breakpt
Value returned is $1 = 180
After the change:
(gdb) backtrace 3
(gdb) finish
Run till exit from #0 second (x=20, y=9) at test.f90:4
0x0000000000404333 in first (num1=16, num2=3) at test.f90:10
10 total = second (num1 + 4, num2 * 3) ! first-breakpt
Value returned is $1 = 180
New test func-trampoline is also added to test this in both backtrace and
finish commands for Intel(R) Fortran Compiler.
2023-08-16 Abdul Basit Ijaz <abdul.b.ijaz@intel.com>
---
gdb/doc/gdb.texinfo | 35 ++++++----
gdb/infcmd.c | 13 ++++
gdb/infrun.c | 16 +++--
gdb/infrun.h | 6 ++
gdb/python/py-frame.c | 15 ++++
gdb/stack.c | 8 +++
gdb/symtab.c | 12 ++++
gdb/symtab.h | 5 ++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++++++++++
.../gdb.fortran/mixed-lang-stack.exp | 10 ++-
11 files changed, 209 insertions(+), 19 deletions(-)
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
index abebad04632..c96fc0ca688 100644
--- a/gdb/doc/gdb.texinfo
+++ b/gdb/doc/gdb.texinfo
@@ -6401,23 +6401,34 @@ so-called @dfn{trampoline functions}, which wrap the actual function call
(the target of the trampoline). The compiler might mark such a trampoline in
its debug information. Often, such trampolines do not have any source line
information associated with them which will lead the @code{step} command to
-behave like a @code{next} and skip the function call completely.
+behave like a @code{next} and skip the function call completely. Similarly,
+the @code{finish} command will return to some trampoline frame entry instead
+of returning to the call site.
The default is @code{set skip-trampoline-functions on} and it will cause the
-@code{step} command to treat these trampolines differently. When issuing a
-@code{step} at the call site of a trampoline function if
-@code{skip-trampoline-functions} is set @value{GDBN} will attempt to determine
+@code{step} and the @code{finish} command to treat these trampolines
+differently.
+
+When issuing a @code{step} at the call site of a trampoline function, if
+@code{skip-trampoline-functions} is set, @value{GDBN} will attempt to determine
the target of the trampoline and then step through the trampoline stopping at
the target. If the target could not be found or was not given in the debug
info, @value{GDBN} will simply continue execution until it leaves the
-trampoline code again, even if the trampoline has no line info associated with
-it. When returning from a target function call and stepping back into the
-trampoline, @value{GDBN} will again step through the trampoline towards the
-call site. Additionally, even if stopped in a trampoline function with source
-line information, issuing a @code{step} will prompt @value{GDBN} to resume
-execution until leaving the trampoline region again. The @code{stepi} command
-is not affected by the setting which is enabled by default. Currently, only
-DWARF trampolines marked via DW_AT_trampoline are supported by this.
+trampoline code again, even if the trampoline has no line info associated
+with it. When returning from a target function call and stepping back into
+the trampoline, @value{GDBN} will again step through the trampoline towards
+the call site. Additionally, even if stopped in a trampoline function with
+source line information, issuing a @code{step} will prompt @value{GDBN} to
+resume execution until leaving the trampoline region again. The @code{stepi}
+command is not affected by the setting which is enabled by default.
+
+When issuing a @code{finish} command at the target of a trampoline function, if
+@code{skip-trampoline-functions} is set, @value{GDBN} will ignore all the
+trampoline frames and will reach the first non-trampoline frame to return to
+the call site of the current function.
+
+Currently, only DWARF trampolines marked via DW_AT_trampoline are supported by
+this.
@item set skip-trampoline-functions off
Causes the @code{step} command to completely ignore any trampoline information
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index 15702f84894..793db6f8566 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1853,6 +1853,19 @@ finish_command (const char *arg, int from_tty)
if (frame == 0)
error (_("\"finish\" not meaningful in the outermost frame."));
+ if (skip_trampoline_functions)
+ {
+ for (int i = 0;
+ (i < MAX_TRAMPOLINE_CHAIN_SIZE && (frame != nullptr)
+ && in_trampoline_frame (frame)); ++i)
+ frame = get_prev_frame (frame);
+
+ if (frame == nullptr)
+ error (_("\"finish\" not meaningful in the outermost non-trampoline \
+frame. Consider running \"set skip-trampoline-functions off\", to stop in \
+trampoline frames for the \"finish\" command."));
+ }
+
clear_proceed_status (0);
tp = inferior_thread ();
diff --git a/gdb/infrun.c b/gdb/infrun.c
index b2597681e00..8d193286e81 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -188,7 +188,7 @@ static bool detach_fork = true;
DW_AT_trampoline by the compiler. If false, GDB will ignore the
attribute. */
-static bool skip_trampoline_functions = true;
+bool skip_trampoline_functions = true;
bool debug_infrun = false;
static void
@@ -7572,6 +7572,14 @@ process_event_stop_test (struct execution_control_state *ecs)
keep_going (ecs);
return;
}
+ else if (skip_trampoline_functions && in_trampoline_function (stop_pc))
+ {
+ /* While reverse stepping if we are in a function call trampoline
+ we will just continue single step in the hope of leaving the
+ trampoline again soon. */
+ keep_going (ecs);
+ return;
+ }
}
/* This always returns the sal for the inner-most frame when we are in a
@@ -10224,9 +10232,9 @@ Options are 'forward' or 'reverse'."),
&skip_trampoline_functions, _("\
Set whether gdb attempts to hide trampolines marked in the debug info."), _("\
Show whether gdb attempts to hide trampolines marked in the debug info."), _("\
-If on, while stepping gdb will skip through functions and inlined functions\n\
-marked as trampolines by the compiler. If off, gdb will ignore such function\n\
-trampolines."),
+If on, the step and finish commands will skip through functions and inlined\n\
+functions marked as trampolines by the compiler. If off, gdb will ignore\n\
+such function trampolines."),
nullptr, show_skip_trampoline_functions, &setlist,
&showlist);
diff --git a/gdb/infrun.h b/gdb/infrun.h
index d4a6b7892b5..feb5f4414ec 100644
--- a/gdb/infrun.h
+++ b/gdb/infrun.h
@@ -106,6 +106,12 @@ extern bool non_stop;
starting an inferior. */
extern bool disable_randomization;
+/* If set (default) GDB will step through functions/inlined subroutines marked
+ DW_AT_trampoline by the compiler. If false, GDB will ignore the
+ attribute. */
+
+extern bool skip_trampoline_functions;
+
/* Returns a unique identifier for the current stop. This can be used
to tell whether a command has proceeded the inferior past the
current location. */
diff --git a/gdb/python/py-frame.c b/gdb/python/py-frame.c
index 1a55e514e39..b2dbfd41fc6 100644
--- a/gdb/python/py-frame.c
+++ b/gdb/python/py-frame.c
@@ -71,6 +71,20 @@ frame_object_to_frame_info (PyObject *obj)
if (frame_obj->frame_id_is_next)
frame = get_prev_frame (frame);
+ if (skip_trampoline_functions)
+ {
+ for (int i = 0;
+ (i < MAX_TRAMPOLINE_CHAIN_SIZE && (frame != nullptr)
+ && in_trampoline_frame (frame)); ++i)
+ {
+ frame = get_prev_frame (frame);
+ if (frame == nullptr)
+ error (_("Skipping trampoline frames is not meaningful. Consider \
+running \"set skip-trampoline-functions off\", to stop in trampoline \
+frames."));
+ }
+ }
+
return frame;
}
@@ -662,6 +676,7 @@ gdbpy_selected_frame (PyObject *self, PyObject *args)
return frame_info_to_frame_object (frame);
}
+
/* Implementation of gdb.stop_reason_string (Integer) -> String.
Return a string explaining the unwind stop reason. */
diff --git a/gdb/stack.c b/gdb/stack.c
index 002bf580634..915ac398e09 100644
--- a/gdb/stack.c
+++ b/gdb/stack.c
@@ -2054,6 +2054,14 @@ backtrace_command_1 (const frame_print_options &fp_opts,
{
QUIT;
+ if (in_trampoline_frame (fi))
+ {
+ /* Trampoline frames are not printed so they are not counted in
+ the backtrace limit. */
+ count++;
+ continue;
+ }
+
/* Don't use print_stack_frame; if an error() occurs it probably
means further attempts to backtrace would fail (on the other
hand, perhaps the code does or could be fixed to make sure
diff --git a/gdb/symtab.c b/gdb/symtab.c
index 855f9b41ec7..3ec4ad6dac2 100644
--- a/gdb/symtab.c
+++ b/gdb/symtab.c
@@ -4066,6 +4066,18 @@ in_trampoline_function (CORE_ADDR pc)
/* See symtab.h. */
+bool
+in_trampoline_frame (frame_info_ptr fi)
+{
+ CORE_ADDR pc;
+ if (get_frame_pc_if_available (fi, &pc))
+ return in_trampoline_function (pc);
+
+ return false;
+}
+
+/* See symtab.h. */
+
CORE_ADDR
find_function_trampoline_target (CORE_ADDR pc)
{
diff --git a/gdb/symtab.h b/gdb/symtab.h
index fc2d82e6567..f02b3840f00 100644
--- a/gdb/symtab.h
+++ b/gdb/symtab.h
@@ -2317,6 +2317,11 @@ extern CORE_ADDR find_solib_trampoline_target (frame_info_ptr, CORE_ADDR);
extern bool in_trampoline_function (CORE_ADDR pc);
+/* Return whether or not the pc of current frame is within a block that belongs
+ to a function that is marked as a trampoline by the compiler. */
+
+extern bool in_trampoline_frame (frame_info_ptr);
+
/* Find the target of a trampoline function marked via the DW_AT_trampoline
attribute and return its address. Returns 0 if the pc is not contained
in a trampoline function (inlined or not). If DW_AT_trampoline
diff --git a/gdb/testsuite/gdb.fortran/func-trampoline.exp b/gdb/testsuite/gdb.fortran/func-trampoline.exp
new file mode 100644
index 00000000000..3bc4d1c0714
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/func-trampoline.exp
@@ -0,0 +1,69 @@
+# Copyright 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/> .
+
+# Test finish and backtrace commands for functions with trampoline
+# calls. Also checks if trampoline frames are filtered while printing
+# stack and finish command returns to real function.
+
+require allow_fortran_tests
+
+if {![test_compiler_info {ifx-*} f90]} {
+ untested "Test is only applicable for IFX where\
+ compiler emits trampoline DIE in Dwarf"
+ return -1
+}
+
+standard_testfile ".f90"
+load_lib fortran.exp
+
+if {[prepare_for_testing "failed to prepare" ${testfile} ${srcfile} \
+ {debug f90}]} {
+ return -1
+}
+
+if {![fortran_runto_main]} {
+ return -1
+}
+
+set inner_loc [gdb_get_line_number "second-breakpt"]
+set middle_loc [gdb_get_line_number "first-breakpt"]
+set outer_loc [gdb_get_line_number "main-outer-loc"]
+set fill "\[^\r\n\]*"
+
+set inner_desc "second \\(x=20, y=9\\) at ${fill}$srcfile:$inner_loc"
+set middle_desc "first \\(num1=16, num2=3\\) at ${fill}$srcfile:$middle_loc"
+set outer_desc ".* at .*$srcfile:$outer_loc"
+
+# Set breakpoint inside the innermost function 'second'.
+gdb_breakpoint "$srcfile:$inner_loc"
+gdb_continue_to_breakpoint "innermost-body" ".*$srcfile:$inner_loc.*"
+
+# Limit the backtrace to 3 frames and ensure both frames for first
+# and second function are shown only and trampoline frames are filtered.
+gdb_test "backtrace 3" [multi_line \
+ "#$decimal.* $inner_desc" \
+ "#$decimal.* $middle_desc" \
+ "#$decimal.* $outer_desc.*"] \
+"backtrace, test trampoline frames are filtered"
+
+# Finish the function calls. Normally we expect to see a "Value
+# returned is ..." line.
+set value_returned "(\r\nValue returned is $valnum_re = 180)"
+
+gdb_test "finish" [multi_line \
+ "Run till exit from #0 $fill second \\(x=20, y=9\\) $fill" \
+ "${fill}first \\(num1=16, num2=3\\)${fill}" \
+ "${fill}${value_returned}"] \
+ "finish the innermost frame"
diff --git a/gdb/testsuite/gdb.fortran/func-trampoline.f90 b/gdb/testsuite/gdb.fortran/func-trampoline.f90
new file mode 100644
index 00000000000..652733b44fd
--- /dev/null
+++ b/gdb/testsuite/gdb.fortran/func-trampoline.f90
@@ -0,0 +1,39 @@
+! Copyright 2023 Free Software Foundation, Inc.
+!
+! This program is free software; you can redistribute it and/or modify
+! it under the terms of the GNU General Public License as published by
+! the Free Software Foundation; either version 3 of the License, or
+! (at your option) any later version.
+!
+! This program is distributed in the hope that it will be useful,
+! but WITHOUT ANY WARRANTY; without even the implied warranty of
+! MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+! GNU General Public License for more details.
+!
+! You should have received a copy of the GNU General Public License
+! along with this program. If not, see <http://www.gnu.org/licenses/> .
+
+! Source code for func-trampoline.exp.
+
+integer(kind=4) function second(x, y)
+ integer(kind=4), intent(in) :: x
+ integer(kind=4), intent(in) :: y
+
+ second = x * y ! second-breakpt
+end function
+
+integer(kind=4) function first(num1, num2)
+ integer(kind=4), intent(in) :: num1
+ integer(kind=4), intent(in) :: num2
+
+ first = second (num1 + 4, num2 * 3) ! first-breakpt
+end function
+
+program func_trampoline
+ integer(kind=4) :: total
+
+ total = first(16, 3) ! main-outer-loc
+
+ write(*,*) "Result is ", total, "\n"
+ ! Expected: 180
+end program func_trampoline
diff --git a/gdb/testsuite/gdb.fortran/mixed-lang-stack.exp b/gdb/testsuite/gdb.fortran/mixed-lang-stack.exp
index 5fc36f7b95b..71a171edcf0 100644
--- a/gdb/testsuite/gdb.fortran/mixed-lang-stack.exp
+++ b/gdb/testsuite/gdb.fortran/mixed-lang-stack.exp
@@ -41,7 +41,7 @@ set have_index [exec_has_index_section $binfile]
# value to pass to GDB's 'set language ...' command.
proc run_tests { lang } {
with_test_prefix "lang=${lang}" {
- global binfile hex have_index
+ global binfile hex have_index decimal
clean_restart ${binfile}
@@ -63,6 +63,10 @@ proc run_tests { lang } {
set e_arg "\['\"\]abcdef\['\"\]"
set 1b_args "\[^\r\n\]+$e_arg\[^\r\n\]+"
set 1g_args "obj=\[^\r\n\]+"
+ # Generic decimal number is checked in regex for Frame #8 and #9 to
+ # handle filtered trampoline frames. Since Frame#8 and Frame#10 are
+ # set to trampoline in DWARF by Intel® Fortran Compiler (ifx), they
+ # are not shown by the backtrace command.
set bt_stack \
[multi_line \
"#0\\s+breakpt \\(\\) at \[^\r\n\]+" \
@@ -73,8 +77,8 @@ proc run_tests { lang } {
"#5\\s+$hex in mixed_func_1d \\(\[^\r\n\]+\\) at \[^\r\n\]+" \
"#6\\s+$hex in mixed_func_1c \\(\[^\r\n\]+\\) at \[^\r\n\]+" \
"#7\\s+$hex in mixed_func_1b \\($1b_args\\) at \[^\r\n\]+" \
- "#8\\s+$hex in mixed_func_1a \\(\\) at \[^\r\n\]+" \
- "#9\\s+$hex in mixed_stack_main \\(\\) at \[^\r\n\]+" ]
+ "#$decimal\\s+$hex in mixed_func_1a \\(\\) at \[^\r\n\]+" \
+ "#$decimal\\s+$hex in mixed_stack_main \\(\\) at \[^\r\n\]+" ]
set main_args "argc=1, argv=${hex}( \[^\r\n\]+)?"
set bt_stack_kfail \
[multi_line \
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
` (3 preceding siblings ...)
2023-08-16 21:53 ` [PATCH v5 4/4] gdb: Skip trampoline frames in the stack for printing or finish command Abdul Basit Ijaz
@ 2023-08-17 5:26 ` Eli Zaretskii
2023-08-17 7:37 ` Ijaz, Abdul B
2023-09-03 19:14 ` [PING][PATCH " Ijaz, Abdul B
5 siblings, 1 reply; 13+ messages in thread
From: Eli Zaretskii @ 2023-08-17 5:26 UTC (permalink / raw)
To: Abdul Basit Ijaz; +Cc: gdb-patches, JiniSusan.George, tom, blarsen
> From: Abdul Basit Ijaz <abdul.b.ijaz@intel.com>
> Cc: abdul.b.ijaz@intel.com,
> JiniSusan.George@amd.com,
> tom@tromey.com,
> eliz@gnu.org,
> blarsen@redhat.com
> Date: Wed, 16 Aug 2023 23:53:45 +0200
>
> Changes since V4:
> * Fixed reverse finish for trampolines.
> * Elide trampoline frames in backtrace for frame-filters case.
> * Use target subroutine address instead of name in case of ref type
> trampoline target.
> * Fixed loops formatting and now split line where possible at ";".
>
> V4 series of patches can be found here:
> https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No changes in documentation since v4, right?
^ permalink raw reply [flat|nested] 13+ messages in thread
* RE: [PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-08-17 5:26 ` [PATCH v5 0/4] GDB support for DW_AT_trampoline Eli Zaretskii
@ 2023-08-17 7:37 ` Ijaz, Abdul B
0 siblings, 0 replies; 13+ messages in thread
From: Ijaz, Abdul B @ 2023-08-17 7:37 UTC (permalink / raw)
To: Eli Zaretskii; +Cc: gdb-patches, JiniSusan.George, tom, blarsen
Hi Eli,
Yes nothing changed in documentation since V4. Thanks
Best Regards
Abdul Basit
-----Original Message-----
From: Eli Zaretskii <eliz@gnu.org>
Sent: Thursday, August 17, 2023 7:27 AM
To: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Cc: gdb-patches@sourceware.org; JiniSusan.George@amd.com; tom@tromey.com; blarsen@redhat.com
Subject: Re: [PATCH v5 0/4] GDB support for DW_AT_trampoline
> From: Abdul Basit Ijaz <abdul.b.ijaz@intel.com>
> Cc: abdul.b.ijaz@intel.com,
> JiniSusan.George@amd.com,
> tom@tromey.com,
> eliz@gnu.org,
> blarsen@redhat.com
> Date: Wed, 16 Aug 2023 23:53:45 +0200
>
> Changes since V4:
> * Fixed reverse finish for trampolines.
> * Elide trampoline frames in backtrace for frame-filters case.
> * Use target subroutine address instead of name in case of ref type
> trampoline target.
> * Fixed loops formatting and now split line where possible at ";".
>
> V4 series of patches can be found here:
> https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No changes in documentation since v4, right?
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
` (4 preceding siblings ...)
2023-08-17 5:26 ` [PATCH v5 0/4] GDB support for DW_AT_trampoline Eli Zaretskii
@ 2023-09-03 19:14 ` Ijaz, Abdul B
2023-09-27 21:05 ` [PING 2][PATCH " Ijaz, Abdul B
5 siblings, 1 reply; 13+ messages in thread
From: Ijaz, Abdul B @ 2023-09-03 19:14 UTC (permalink / raw)
To: gdb-patches; +Cc: JiniSusan.George, tom, eliz, blarsen
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* RE: [PING 2][PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-09-03 19:14 ` [PING][PATCH " Ijaz, Abdul B
@ 2023-09-27 21:05 ` Ijaz, Abdul B
2023-12-28 11:30 ` [PING 3][PATCH " Ijaz, Abdul B
0 siblings, 1 reply; 13+ messages in thread
From: Ijaz, Abdul B @ 2023-09-27 21:05 UTC (permalink / raw)
To: gdb-patches; +Cc: JiniSusan.George, tom, eliz, blarsen
Ping!
Thanks & Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Sunday, September 3, 2023 9:14 PM
To: gdb-patches@sourceware.org
Cc: JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-09-27 21:05 ` [PING 2][PATCH " Ijaz, Abdul B
@ 2023-12-28 11:30 ` Ijaz, Abdul B
2024-01-10 7:40 ` George, Jini Susan
0 siblings, 1 reply; 13+ messages in thread
From: Ijaz, Abdul B @ 2023-12-28 11:30 UTC (permalink / raw)
To: gdb-patches; +Cc: JiniSusan.George, tom, eliz, blarsen
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Wednesday, September 27, 2023 11:05 PM
To: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org>
Cc: 'JiniSusan.George@amd.com' <JiniSusan.George@amd.com>; 'tom@tromey.com' <tom@tromey.com>; 'eliz@gnu.org' <eliz@gnu.org>; 'blarsen@redhat.com' <blarsen@redhat.com>
Subject: RE: [PING 2][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Thanks & Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Sunday, September 3, 2023 9:14 PM
To: gdb-patches@sourceware.org
Cc: JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
2023-12-28 11:30 ` [PING 3][PATCH " Ijaz, Abdul B
@ 2024-01-10 7:40 ` George, Jini Susan
2024-01-10 9:22 ` Ijaz, Abdul B
0 siblings, 1 reply; 13+ messages in thread
From: George, Jini Susan @ 2024-01-10 7:40 UTC (permalink / raw)
To: Ijaz, Abdul B, gdb-patches; +Cc: tom, eliz, blarsen
[-- Attachment #1: Type: text/plain, Size: 6713 bytes --]
[Public]
Hi Abdul,
Thank you for incorporating the comments. A few suggestions/ comments wrt the new functionality of skipping the stack frames with the 'bt' and 'finish' commands below.
* Shouldn't the frame skipping be done for the 'return' command also ?
* Shouldn't the frame skipping be enabled for the MI command '-stack-list-frames' also ?
* The frame skipping does not seem to be working correctly for the 'backtrace -n' command (to list 'n' nbr of the outermost frames).
* As of now, there does not seem to be a way to turn off the filtering of the trampoline frames. Shouldn't there be one ?
* One suggestion for the trampoline frame filtering is to piggyback on the current python frame filtering (plus FrameDecorator) mechanism. This might mean including adding a filter for this to the current source. This would enable users to use the existing commands to turn off frame filtering. But this is just a suggestion, and it would be good to go by what the maintainers say wrt this.
Thanks,
Jini.
________________________________
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Thursday, December 28, 2023 17:00
To: gdb-patches@sourceware.org <gdb-patches@sourceware.org>
Cc: George, Jini Susan <JiniSusan.George@amd.com>; tom@tromey.com <tom@tromey.com>; eliz@gnu.org <eliz@gnu.org>; blarsen@redhat.com <blarsen@redhat.com>
Subject: RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Wednesday, September 27, 2023 11:05 PM
To: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org>
Cc: 'JiniSusan.George@amd.com' <JiniSusan.George@amd.com>; 'tom@tromey.com' <tom@tromey.com>; 'eliz@gnu.org' <eliz@gnu.org>; 'blarsen@redhat.com' <blarsen@redhat.com>
Subject: RE: [PING 2][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Thanks & Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Sunday, September 3, 2023 9:14 PM
To: gdb-patches@sourceware.org
Cc: JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; JiniSusan.George@amd.com; tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de<http://www.intel.de> <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
2024-01-10 7:40 ` George, Jini Susan
@ 2024-01-10 9:22 ` Ijaz, Abdul B
2024-01-10 16:59 ` George, Jini Susan
0 siblings, 1 reply; 13+ messages in thread
From: Ijaz, Abdul B @ 2024-01-10 9:22 UTC (permalink / raw)
To: George, Jini Susan, gdb-patches; +Cc: tom, eliz, blarsen
[-- Attachment #1: Type: text/plain, Size: 10863 bytes --]
Hi Jini,
Thanks a lot for the feedback. I have added my response here :
* Shouldn't the frame skipping be done for the 'return' command also ?
Yes this is missing so I will add to return command also in next V6 series.
* Shouldn't the frame skipping be enabled for the MI command '-stack-list-frames' also ?
Yes this is also missing and need to check this. Will be added to this MI command also in V6 series.
* The frame skipping does not seem to be working correctly for the 'backtrace -n' command (to list 'n' nbr of the outermost frames).
In commit message I think due to “#” character at beginning backtrace lines are removed from commit message. I will update the commit message to show the lines or have you seen some issue as otherwise it works fine. “Backtrace n” shows like following output without trampoline frames. Missing #1 & #3 here are trampoline frames :
(gdb) bt 3
#0 second (x=20, y=9) at trampoline.f90:4
#2 0x00000000004051e3 in first (num1=16, num2=3) at trampoline.f90:10
#4 0x0000000000405234 in func_trampoline () at trampoline.f90:17
(gdb) show skip-trampoline-functions
Skipping trampoline functions is on.
(gdb) set skip-trampoline-functions off
(gdb) bt 3
#0 second (x=20, y=9) at trampoline.f90:4
#1 0x0000000000405209 in second_.t74p.t75p () at trampoline.f90:12
#2 0x00000000004051e3 in first (num1=16, num2=3) at trampoline.f90:10
(More stack frames follow...)
* As of now, there does not seem to be a way to turn off the filtering of the trampoline frames. Shouldn't there be one ?
There is skip-trampoline-functions option added in patch #4 of the series. So it should be possible to turn it off. Would not it be sufficient? Though found out that its handling for backtrace command is missing which I will add like shown above when I will send next V6 series of patch for review.
* One suggestion for the trampoline frame filtering is to piggyback on the current python frame filtering (plus FrameDecorator) mechanism. This might mean including adding a filter for this to the current source. This would enable users to use the existing commands to turn off frame filtering. But this is just a suggestion, and it would be good to go by what the maintainers say wrt this
Not much familiar with it but as you said need to see maintainers feedback regarding this and then may investigate accordingly. What is done so far regarding python frame filters is that now handling filter trampoline frames also using “skip-trampoline-functions” configure option. By default python frame filters will not show trampoline frames but can be enabled to show this by using this option.
Thanks & Best Regards
Abdul Basit
From: George, Jini Susan <JiniSusan.George@amd.com>
Sent: Wednesday, January 10, 2024 8:41 AM
To: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; gdb-patches@sourceware.org
Cc: tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: Re: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
[Public]
Hi Abdul,
Thank you for incorporating the comments. A few suggestions/ comments wrt the new functionality of skipping the stack frames with the 'bt' and 'finish' commands below.
* Shouldn't the frame skipping be done for the 'return' command also ?
* Shouldn't the frame skipping be enabled for the MI command '-stack-list-frames' also ?
* The frame skipping does not seem to be working correctly for the 'backtrace -n' command (to list 'n' nbr of the outermost frames).
* As of now, there does not seem to be a way to turn off the filtering of the trampoline frames. Shouldn't there be one ?
* One suggestion for the trampoline frame filtering is to piggyback on the current python frame filtering (plus FrameDecorator) mechanism. This might mean including adding a filter for this to the current source. This would enable users to use the existing commands to turn off frame filtering. But this is just a suggestion, and it would be good to go by what the maintainers say wrt this.
Thanks,
Jini.
________________________________
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Sent: Thursday, December 28, 2023 17:00
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org> <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: George, Jini Susan <JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>>; tom@tromey.com<mailto:tom@tromey.com> <tom@tromey.com<mailto:tom@tromey.com>>; eliz@gnu.org<mailto:eliz@gnu.org> <eliz@gnu.org<mailto:eliz@gnu.org>>; blarsen@redhat.com<mailto:blarsen@redhat.com> <blarsen@redhat.com<mailto:blarsen@redhat.com>>
Subject: RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Wednesday, September 27, 2023 11:05 PM
To: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: 'JiniSusan.George@amd.com' <JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>>; 'tom@tromey.com' <tom@tromey.com<mailto:tom@tromey.com>>; 'eliz@gnu.org' <eliz@gnu.org<mailto:eliz@gnu.org>>; 'blarsen@redhat.com' <blarsen@redhat.com<mailto:blarsen@redhat.com>>
Subject: RE: [PING 2][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Thanks & Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Sunday, September 3, 2023 9:14 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Cc: JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>; tom@tromey.com<mailto:tom@tromey.com>; eliz@gnu.org<mailto:eliz@gnu.org>; blarsen@redhat.com<mailto:blarsen@redhat.com>
Subject: RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>; JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>; tom@tromey.com<mailto:tom@tromey.com>; eliz@gnu.org<mailto:eliz@gnu.org>; blarsen@redhat.com<mailto:blarsen@redhat.com>
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de<http://www.intel.de> <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
2024-01-10 9:22 ` Ijaz, Abdul B
@ 2024-01-10 16:59 ` George, Jini Susan
0 siblings, 0 replies; 13+ messages in thread
From: George, Jini Susan @ 2024-01-10 16:59 UTC (permalink / raw)
To: Ijaz, Abdul B, gdb-patches; +Cc: tom, eliz, blarsen
[-- Attachment #1: Type: text/plain, Size: 11767 bytes --]
[Public]
Hi Abdul,
Wrt the backtrace command, I was mentioning the 'backtrace -n' command, with a minus before the 'n', so that 'n' number of the outermost frames, rather than the innermost ones are displayed. That did not seem to work correctlly.
Thanks,
Jini.
________________________________
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com>
Sent: Wednesday, January 10, 2024 14:52
To: George, Jini Susan <JiniSusan.George@amd.com>; gdb-patches@sourceware.org <gdb-patches@sourceware.org>
Cc: tom@tromey.com <tom@tromey.com>; eliz@gnu.org <eliz@gnu.org>; blarsen@redhat.com <blarsen@redhat.com>
Subject: RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
[Public]
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
Hi Jini,
Thanks a lot for the feedback. I have added my response here :
* Shouldn't the frame skipping be done for the 'return' command also ?
Yes this is missing so I will add to return command also in next V6 series.
* Shouldn't the frame skipping be enabled for the MI command '-stack-list-frames' also ?
Yes this is also missing and need to check this. Will be added to this MI command also in V6 series.
* The frame skipping does not seem to be working correctly for the 'backtrace -n' command (to list 'n' nbr of the outermost frames).
In commit message I think due to “#” character at beginning backtrace lines are removed from commit message. I will update the commit message to show the lines or have you seen some issue as otherwise it works fine. “Backtrace n” shows like following output without trampoline frames. Missing #1 & #3 here are trampoline frames :
(gdb) bt 3
#0 second (x=20, y=9) at trampoline.f90:4
#2 0x00000000004051e3 in first (num1=16, num2=3) at trampoline.f90:10
#4 0x0000000000405234 in func_trampoline () at trampoline.f90:17
(gdb) show skip-trampoline-functions
Skipping trampoline functions is on.
(gdb) set skip-trampoline-functions off
(gdb) bt 3
#0 second (x=20, y=9) at trampoline.f90:4
#1 0x0000000000405209 in second_.t74p.t75p () at trampoline.f90:12
#2 0x00000000004051e3 in first (num1=16, num2=3) at trampoline.f90:10
(More stack frames follow...)
* As of now, there does not seem to be a way to turn off the filtering of the trampoline frames. Shouldn't there be one ?
There is skip-trampoline-functions option added in patch #4 of the series. So it should be possible to turn it off. Would not it be sufficient? Though found out that its handling for backtrace command is missing which I will add like shown above when I will send next V6 series of patch for review.
* One suggestion for the trampoline frame filtering is to piggyback on the current python frame filtering (plus FrameDecorator) mechanism. This might mean including adding a filter for this to the current source. This would enable users to use the existing commands to turn off frame filtering. But this is just a suggestion, and it would be good to go by what the maintainers say wrt this
Not much familiar with it but as you said need to see maintainers feedback regarding this and then may investigate accordingly. What is done so far regarding python frame filters is that now handling filter trampoline frames also using “skip-trampoline-functions” configure option. By default python frame filters will not show trampoline frames but can be enabled to show this by using this option.
Thanks & Best Regards
Abdul Basit
From: George, Jini Susan <JiniSusan.George@amd.com>
Sent: Wednesday, January 10, 2024 8:41 AM
To: Ijaz, Abdul B <abdul.b.ijaz@intel.com>; gdb-patches@sourceware.org
Cc: tom@tromey.com; eliz@gnu.org; blarsen@redhat.com
Subject: Re: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
[Public]
Hi Abdul,
Thank you for incorporating the comments. A few suggestions/ comments wrt the new functionality of skipping the stack frames with the 'bt' and 'finish' commands below.
* Shouldn't the frame skipping be done for the 'return' command also ?
* Shouldn't the frame skipping be enabled for the MI command '-stack-list-frames' also ?
* The frame skipping does not seem to be working correctly for the 'backtrace -n' command (to list 'n' nbr of the outermost frames).
* As of now, there does not seem to be a way to turn off the filtering of the trampoline frames. Shouldn't there be one ?
* One suggestion for the trampoline frame filtering is to piggyback on the current python frame filtering (plus FrameDecorator) mechanism. This might mean including adding a filter for this to the current source. This would enable users to use the existing commands to turn off frame filtering. But this is just a suggestion, and it would be good to go by what the maintainers say wrt this.
Thanks,
Jini.
________________________________
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Sent: Thursday, December 28, 2023 17:00
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org> <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: George, Jini Susan <JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>>; tom@tromey.com<mailto:tom@tromey.com> <tom@tromey.com<mailto:tom@tromey.com>>; eliz@gnu.org<mailto:eliz@gnu.org> <eliz@gnu.org<mailto:eliz@gnu.org>>; blarsen@redhat.com<mailto:blarsen@redhat.com> <blarsen@redhat.com<mailto:blarsen@redhat.com>>
Subject: RE: [PING 3][PATCH v5 0/4] GDB support for DW_AT_trampoline
Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Wednesday, September 27, 2023 11:05 PM
To: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: 'JiniSusan.George@amd.com' <JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>>; 'tom@tromey.com' <tom@tromey.com<mailto:tom@tromey.com>>; 'eliz@gnu.org' <eliz@gnu.org<mailto:eliz@gnu.org>>; 'blarsen@redhat.com' <blarsen@redhat.com<mailto:blarsen@redhat.com>>
Subject: RE: [PING 2][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Thanks & Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B
Sent: Sunday, September 3, 2023 9:14 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Cc: JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>; tom@tromey.com<mailto:tom@tromey.com>; eliz@gnu.org<mailto:eliz@gnu.org>; blarsen@redhat.com<mailto:blarsen@redhat.com>
Subject: RE: [PING][PATCH v5 0/4] GDB support for DW_AT_trampoline
Ping!
Best Regards
Abdul Basit
-----Original Message-----
From: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Sent: Wednesday, August 16, 2023 11:54 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Cc: Ijaz, Abdul B <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>; JiniSusan.George@amd.com<mailto:JiniSusan.George@amd.com>; tom@tromey.com<mailto:tom@tromey.com>; eliz@gnu.org<mailto:eliz@gnu.org>; blarsen@redhat.com<mailto:blarsen@redhat.com>
Subject: [PATCH v5 0/4] GDB support for DW_AT_trampoline
From: "Ijaz, Abdul B" <abdul.b.ijaz@intel.com<mailto:abdul.b.ijaz@intel.com>>
Hi,
please find attached a series of patches that attempts to enable GDB to handle the DW_AT_trampoline attribute on functions and inlined functions.
DW_AT_trampoline can be emitted by the compiler for functions that are compiler generated trampolines (e.g. wrapping other function calls).
In doing so, the compiler can specify a target as the value of DW_AT_trampoline which the trampoline is wrapping.
This series enables GDB to recognize the DW_AT_trampoline and store the target for a given function. Patch 3 adapts GDB's stepping behavior when dealing with trampolines and attempts to - by default - hide these trampolines from the user. When about to step into a trampoline, instead, GDB will try and step through the trampoline and directly towards the target. Patch 4 in V2 now skip trampoline frames in the stack for printing or finish command. A new setting has been introduced to turn off this modified stepping behavior or printing of stack for trampolines.
The motivation for these patches comes from ifx which emits the attribute for some of its compiler generated functions. As I do not know of any other compiler (especially gcc/gfortran) emitting DW_AT_trampoline, I added a gdb.dwarf2 test to this series in order to test the trampoline handling within GDB.
Changes since V4:
* Fixed reverse finish for trampolines.
* Elide trampoline frames in backtrace for frame-filters case.
* Use target subroutine address instead of name in case of ref type
trampoline target.
* Fixed loops formatting and now split line where possible at ";".
V4 series of patches can be found here:
https://sourceware.org/pipermail/gdb-patches/2023-August/201263.html
No regression seen on testing x64/native-extended-gdbserver(-m32)/x32/
native-gdbserver test configurations for these changes using Ubuntu 22.
Thanks & Best Regards
Abdul Basit
Abdul Basit Ijaz (1):
gdb: Skip trampoline frames in the stack for printing or finish
command.
Nils-Christian Kempke (3):
gdb, dwarf: add support for DW_AT_trampoline in DWARF reader
gdb/symtab: add lookup for trampoline functions
gdb/infrun: handle stepping through functions with DW_AT_trampoline
gdb/NEWS | 15 ++
gdb/doc/gdb.texinfo | 47 ++++
gdb/dwarf2/read.c | 46 +++-
gdb/gdbtypes.c | 34 ++-
gdb/gdbtypes.h | 113 +++++++-
gdb/infcmd.c | 13 +
gdb/infrun.c | 93 ++++++-
gdb/infrun.h | 10 +
gdb/python/py-frame.c | 15 ++
gdb/stack.c | 8 +
gdb/symtab.c | 80 ++++++
gdb/symtab.h | 19 ++
.../gdb.dwarf2/dw2-function-trampolines.c | 80 ++++++
.../gdb.dwarf2/dw2-function-trampolines.exp | 245 ++++++++++++++++++
gdb/testsuite/gdb.fortran/func-trampoline.exp | 69 +++++
gdb/testsuite/gdb.fortran/func-trampoline.f90 | 39 +++
.../gdb.fortran/mixed-lang-stack.exp | 10 +-
17 files changed, 922 insertions(+), 14 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.c
create mode 100644 gdb/testsuite/gdb.dwarf2/dw2-function-trampolines.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.exp
create mode 100644 gdb/testsuite/gdb.fortran/func-trampoline.f90
--
2.34.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de<http://www.intel.de> <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de<http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2024-01-10 16:59 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-08-16 21:53 [PATCH v5 0/4] GDB support for DW_AT_trampoline Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 1/4] gdb, dwarf: add support for DW_AT_trampoline in DWARF reader Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 2/4] gdb/symtab: add lookup for trampoline functions Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 3/4] gdb/infrun: handle stepping through functions with DW_AT_trampoline Abdul Basit Ijaz
2023-08-16 21:53 ` [PATCH v5 4/4] gdb: Skip trampoline frames in the stack for printing or finish command Abdul Basit Ijaz
2023-08-17 5:26 ` [PATCH v5 0/4] GDB support for DW_AT_trampoline Eli Zaretskii
2023-08-17 7:37 ` Ijaz, Abdul B
2023-09-03 19:14 ` [PING][PATCH " Ijaz, Abdul B
2023-09-27 21:05 ` [PING 2][PATCH " Ijaz, Abdul B
2023-12-28 11:30 ` [PING 3][PATCH " Ijaz, Abdul B
2024-01-10 7:40 ` George, Jini Susan
2024-01-10 9:22 ` Ijaz, Abdul B
2024-01-10 16:59 ` George, Jini Susan
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).