[PATCH 0/2] Fix gdb.arch/amd64-gs_base.exp test on FreeBSD

[PATCH 0/2] Fix gdb.arch/amd64-gs_base.exp test on FreeBSD

[John Baldwin]

The gdb.arch/amd64-gs_base.exp test was failing on FreeBSD as it had
some Linux-specific assumptions.  While looking at this I also found
that Linux/amd64 treats the fs_base and gs_base as "system" registers,
but FreeBSD did not.  I originally started by moving the logic from
Linux/amd64 for these registers into i386_register_reggroup_p so it
would be consistent across all OS ABIs, but I think it would be better
to treat these registers as general registers instead which is what
patch 2 does.

John Baldwin (2):
  gdb.arch/amd64-gs_base.exp: Support non-Linux.
  amd64-linux-tdep: Don't treat fs_base and gs_base as system registers.

 gdb/amd64-linux-tdep.c                   |  4 +---
 gdb/testsuite/gdb.arch/amd64-gs_base.exp | 10 ++++------
 2 files changed, 5 insertions(+), 9 deletions(-)

-- 
2.38.1

[PATCH 1/2] gdb.arch/amd64-gs_base.exp: Support non-Linux.

[John Baldwin]

The orig_rax pseudo-register is Linux-specific and isn't relevant to
this test.  The fs_base and gs_base registers are also not treated as
system registers in other OS ABIs.  This allows the test to pass on
FreeBSD.
---
 gdb/testsuite/gdb.arch/amd64-gs_base.exp | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/gdb/testsuite/gdb.arch/amd64-gs_base.exp b/gdb/testsuite/gdb.arch/amd64-gs_base.exp
index a60e1dbe7f4..083c7c9fe1f 100644
--- a/gdb/testsuite/gdb.arch/amd64-gs_base.exp
+++ b/gdb/testsuite/gdb.arch/amd64-gs_base.exp
@@ -32,14 +32,12 @@ gdb_test "print /x \$gs_base" "= $hex" "print gs_base"
 gdb_test "print \$fs_base = 2" "= 2" "set fs_base"
 gdb_test "print \$gs_base = 3" "= 3" "set gs_base"
 
-# Test the presence of fs_base and gs_base on the system
-# register group and values.
+# Test the fs_base and gs_base values.
 #
 set ws "\[\t \]+"
-set info_reg_out [multi_line "info register sys" \
+set info_reg_out [multi_line "info register fs_base gs_base" \
           "fs_base${ws}0x2${ws}2"\
-          "gs_base${ws}0x3${ws}3"\
-          "orig_rax${ws}$hex${ws}\[-\]$decimal" ]
+          "gs_base${ws}0x3${ws}3" ]
 
-gdb_test "info register sys" $info_reg_out\
+gdb_test "info register fs_base gs_base" $info_reg_out\
    "info registers fs_base and gs_base with value"
-- 
2.38.1

[PATCH 2/2] amd64-linux-tdep: Don't treat fs_base and gs_base as system registers.

[John Baldwin]

These registers can be changed directly in userspace, and similar
registers to support TLS on other architectures (tpidr* on ARM and
AArch64, tp on RISC-V) are treated as general purpose registers.
---
 gdb/amd64-linux-tdep.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/gdb/amd64-linux-tdep.c b/gdb/amd64-linux-tdep.c
index e9181649ec8..cbbac1a0c64 100644
--- a/gdb/amd64-linux-tdep.c
+++ b/gdb/amd64-linux-tdep.c
@@ -281,9 +281,7 @@ static int
 amd64_linux_register_reggroup_p (struct gdbarch *gdbarch, int regnum,
 				 const struct reggroup *group)
 { 
-  if (regnum == AMD64_LINUX_ORIG_RAX_REGNUM
-      || regnum == AMD64_FSBASE_REGNUM
-      || regnum == AMD64_GSBASE_REGNUM)
+  if (regnum == AMD64_LINUX_ORIG_RAX_REGNUM)
     return (group == system_reggroup
 	    || group == save_reggroup
 	    || group == restore_reggroup);
-- 
2.38.1

[PING] [PATCH 0/2] Fix gdb.arch/amd64-gs_base.exp test on FreeBSD

[John Baldwin]

On 2/17/23 12:02 PM, John Baldwin wrote:
> The gdb.arch/amd64-gs_base.exp test was failing on FreeBSD as it had
> some Linux-specific assumptions.  While looking at this I also found
> that Linux/amd64 treats the fs_base and gs_base as "system" registers,
> but FreeBSD did not.  I originally started by moving the logic from
> Linux/amd64 for these registers into i386_register_reggroup_p so it
> would be consistent across all OS ABIs, but I think it would be better
> to treat these registers as general registers instead which is what
> patch 2 does.
> 
> John Baldwin (2):
>    gdb.arch/amd64-gs_base.exp: Support non-Linux.
>    amd64-linux-tdep: Don't treat fs_base and gs_base as system registers.
> 
>   gdb/amd64-linux-tdep.c                   |  4 +---
>   gdb/testsuite/gdb.arch/amd64-gs_base.exp | 10 ++++------
>   2 files changed, 5 insertions(+), 9 deletions(-)

Ping.

-- 
John Baldwin

Re: [PING] [PATCH 0/2] Fix gdb.arch/amd64-gs_base.exp test on FreeBSD

[Tom Tromey]

>>>>> "John" == John Baldwin <jhb@FreeBSD.org> writes:

John> On 2/17/23 12:02 PM, John Baldwin wrote:
>> The gdb.arch/amd64-gs_base.exp test was failing on FreeBSD as it had
>> some Linux-specific assumptions.  While looking at this I also found
>> that Linux/amd64 treats the fs_base and gs_base as "system" registers,
>> but FreeBSD did not.  I originally started by moving the logic from
>> Linux/amd64 for these registers into i386_register_reggroup_p so it
>> would be consistent across all OS ABIs, but I think it would be better
>> to treat these registers as general registers instead which is what
>> patch 2 does.
>> John Baldwin (2):
>> gdb.arch/amd64-gs_base.exp: Support non-Linux.
>> amd64-linux-tdep: Don't treat fs_base and gs_base as system registers.
>> gdb/amd64-linux-tdep.c                   |  4 +---
>> gdb/testsuite/gdb.arch/amd64-gs_base.exp | 10 ++++------
>> 2 files changed, 5 insertions(+), 9 deletions(-)

John> Ping.

These make sense to me.  Thank you.

Reviewed-By: Tom Tromey <tom@tromey.com>

Tom