public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Tom de Vries <tdevries@suse.de>
To: gdb-patches@sourceware.org
Subject: [PING][PATCH] [gdb/symtab] Work around PR gas/29517
Date: Mon, 16 Oct 2023 11:21:35 +0200	[thread overview]
Message-ID: <786e514c-6ff7-452e-a52e-5d727942d1c0@suse.de> (raw)
In-Reply-To: <20230928160748.16042-1-tdevries@suse.de>

On 9/28/23 18:07, Tom de Vries via Gdb-patches wrote:
> When using glibc debuginfo generated with gas 2.39, we run into PR gas/29517:
> ...
> $ gdb -q -batch a.out -ex start -ex "p (char *)strstr (\"haha\", \"ah\")"
> Temporary breakpoint 1 at 0x40051b: file hello.c, line 6.
> 
> Temporary breakpoint 1, main () at hello.c:6
> 6	  printf ("hello\n");
> Invalid cast.
> ...
> while without glibc debuginfo installed we get the expected result:
> ...
> $n = 0x7ffff7daa1b1 "aha"
> ...
> and likewise with glibc debuginfo generated with gas 2.40.
> 
> The strstr ifunc resolves to __strstr_sse2_unaligned.  The problem is that gas
> generates dwarf that states that the return type is void:
> ...
> <1><3e1e58>: Abbrev Number: 2 (DW_TAG_subprogram)
>      <3e1e59>   DW_AT_name        : __strstr_sse2_unaligned
>      <3e1e5d>   DW_AT_external    : 1
>      <3e1e5e>   DW_AT_low_pc      : 0xbbd2e
>      <3e1e66>   DW_AT_high_pc     : 0xbc1c3
> ...
> while the return type should be a DW_TAG_unspecified_type, as is the case
> with gas 2.40.
> 
> We can still use the workaround of casting to another function type for both
> __strstr_sse2_unaligned:
> ...
> (gdb) p ((char * (*) (const char *, const char *))__strstr_sse2_unaligned) \
>    ("haha", "ah")
> $n = 0x7ffff7daa211 "aha"
> ...
> and strstr (which requires using *strstr to dereference the ifunc before we
> cast):
> ...
> gdb) p ((char * (*) (const char *, const char *))*strstr) ("haha", "ah")
> $n = 0x7ffff7daa251 "aha"
> ...
> but that's a bit cumbersome to use.
> 
> Work around this in the dwarf reader, such that we have instead:
> ...
> (gdb) p (char *)strstr ("haha", "ah")
> $n = 0x7ffff7daa1b1 "aha"
> ...
> 
> This also requires fixing producer_is_gcc to stop returning true for
> producer "GNU AS 2.39.0".
> 
> Tested on x86_64-linux.
> 

Ping.

Thanks,
- Tom

> PR symtab/30911
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30911
> ---
>   gdb/dwarf2/cu.c                               |  1 +
>   gdb/dwarf2/cu.h                               |  1 +
>   gdb/dwarf2/read.c                             | 23 ++++++++++++
>   gdb/producer.c                                |  8 ++++-
>   .../gdb.dwarf2/dw2-unspecified-type.c         |  9 ++++-
>   .../gdb.dwarf2/dw2-unspecified-type.exp       | 36 +++++++++++++++----
>   6 files changed, 69 insertions(+), 9 deletions(-)
> 
> diff --git a/gdb/dwarf2/cu.c b/gdb/dwarf2/cu.c
> index 89de40daab0..a908ec908cd 100644
> --- a/gdb/dwarf2/cu.c
> +++ b/gdb/dwarf2/cu.c
> @@ -40,6 +40,7 @@ dwarf2_cu::dwarf2_cu (dwarf2_per_cu_data *per_cu,
>       producer_is_icc_lt_14 (false),
>       producer_is_codewarrior (false),
>       producer_is_clang (false),
> +    producer_is_gas_2_39 (false),
>       processing_has_namespace_info (false),
>       load_all_dies (false)
>   {
> diff --git a/gdb/dwarf2/cu.h b/gdb/dwarf2/cu.h
> index 0c15d8b02db..6c611710503 100644
> --- a/gdb/dwarf2/cu.h
> +++ b/gdb/dwarf2/cu.h
> @@ -265,6 +265,7 @@ struct dwarf2_cu
>     bool producer_is_icc_lt_14 : 1;
>     bool producer_is_codewarrior : 1;
>     bool producer_is_clang : 1;
> +  bool producer_is_gas_2_39 : 1;
>   
>     /* When true, the file that we're processing is known to have
>        debugging info for C++ namespaces.  GCC 3.3.x did not produce
> diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
> index 5bbc8e24cf9..bd8cdad6e65 100644
> --- a/gdb/dwarf2/read.c
> +++ b/gdb/dwarf2/read.c
> @@ -11426,6 +11426,8 @@ check_producer (struct dwarf2_cu *cu)
>       cu->producer_is_codewarrior = true;
>     else if (producer_is_clang (cu->producer, &major, &minor))
>       cu->producer_is_clang = true;
> +  else if (startswith (cu->producer, "GNU AS 2.39.0"))
> +    cu->producer_is_gas_2_39 = true;
>     else
>       {
>         /* For other non-GCC compilers, expect their behavior is DWARF version
> @@ -11461,6 +11463,15 @@ producer_is_codewarrior (struct dwarf2_cu *cu)
>     return cu->producer_is_codewarrior;
>   }
>   
> +static bool
> +producer_is_gas_2_39 (struct dwarf2_cu *cu)
> +{
> +  if (!cu->checked_producer)
> +    check_producer (cu);
> +
> +  return cu->producer_is_gas_2_39;
> +}
> +
>   /* Return the accessibility of DIE, as given by DW_AT_accessibility.
>      If that attribute is not available, return the appropriate
>      default.  */
> @@ -14635,6 +14646,18 @@ read_subroutine_type (struct die_info *die, struct dwarf2_cu *cu)
>   
>     type = die_type (die, cu);
>   
> +  if (type->code () == TYPE_CODE_VOID
> +      && !type->is_stub ()
> +      && die->child == nullptr
> +      && producer_is_gas_2_39 (cu))
> +    {
> +      /* Work around PR gas/29517, pretend we have an DW_TAG_unspecified_type
> +	 return type.  */
> +      type = (type_allocator (cu->per_objfile->objfile, cu->lang ())
> +	      .new_type (TYPE_CODE_VOID, 0, nullptr));
> +      type->set_is_stub (true);
> +    }
> +
>     /* The die_type call above may have already set the type for this DIE.  */
>     ftype = get_die_type (die, cu);
>     if (ftype)
> diff --git a/gdb/producer.c b/gdb/producer.c
> index 655eb971283..9fcf749e3d4 100644
> --- a/gdb/producer.c
> +++ b/gdb/producer.c
> @@ -54,13 +54,19 @@ producer_is_gcc (const char *producer, int *major, int *minor)
>         if (minor == NULL)
>   	minor = &min;
>   
> +      /* Skip GNU.  */
> +      cs = &producer[strlen ("GNU ")];
> +
> +      /* Bail out for GNU AS.  */
> +      if (startswith (cs, "AS "))
> +	return 0;
> +
>         /* Skip any identifier after "GNU " - such as "C11" "C++" or "Java".
>   	 A full producer string might look like:
>   	 "GNU C 4.7.2"
>   	 "GNU Fortran 4.8.2 20140120 (Red Hat 4.8.2-16) -mtune=generic ..."
>   	 "GNU C++14 5.0.0 20150123 (experimental)"
>         */
> -      cs = &producer[strlen ("GNU ")];
>         while (*cs && !isspace (*cs))
>   	cs++;
>         if (*cs && isspace (*cs))
> diff --git a/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.c b/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.c
> index 1df09214d4a..e07d9517ff2 100644
> --- a/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.c
> +++ b/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.c
> @@ -17,9 +17,16 @@
>   
>   extern int foo (void);
>   
> +int
> +bar (void)
> +{
> +  asm ("bar_label: .globl bar_label");
> +  return 0;
> +}
> +
>   int
>   main (void)
>   {
> -  int res = foo ();
> +  int res = foo () + bar ();
>     return res;
>   }
> diff --git a/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.exp b/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.exp
> index 5fa6ee544b2..a6f2a57e33e 100644
> --- a/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.exp
> +++ b/gdb/testsuite/gdb.dwarf2/dw2-unspecified-type.exp
> @@ -27,10 +27,18 @@ lassign $foo_res \
>       foo_start foo_len
>   set foo_end "$foo_start + $foo_len"
>   
> +set bar_res \
> +    [function_range bar \
> +	 [list ${srcdir}/${subdir}/$srcfile ${srcdir}/${subdir}/$srcfile2]]
> +lassign $bar_res \
> +    bar_start bar_len
> +set bar_end "$bar_start + $bar_len"
> +
>   # Create the DWARF.
>   set asm_file [standard_output_file $srcfile3]
>   Dwarf::assemble $asm_file {
>       global foo_start foo_end
> +    global bar_start bar_end
>       declare_labels unspecified_type_label
>   
>       cu {} {
> @@ -45,7 +53,19 @@ Dwarf::assemble $asm_file {
>   		{high_pc $foo_end addr}
>   		{type :$unspecified_type_label}
>   	    }
> +	}
> +    }
>   
> +    cu {} {
> +	compile_unit {
> +	    {language @DW_LANG_Mips_Assembler}
> +	    {producer "GNU AS 2.39.0"}
> +	} {
> +	    DW_TAG_subprogram {
> +		{name bar}
> +		{low_pc $bar_start addr}
> +		{high_pc $bar_end addr}
> +	    }
>   	}
>       }
>   }
> @@ -59,12 +79,14 @@ if ![runto_main] {
>       return -1
>   }
>   
> -# Print the function type.  Return type should be stub type, which is printed
> -# as void.
> -gdb_test "ptype foo" "type = void \\(void\\)"
> +foreach f {foo bar} {
> +    # Print the function type.  Return type should be stub type, which is printed
> +    # as void.
> +    gdb_test "ptype $f" "type = void \\(void\\)"
>   
> -# Call the function, casting the function to the correct function type.
> -gdb_test "p ((int (*) ()) foo) ()" " = 0"
> +    # Call the function, casting the function to the correct function type.
> +    gdb_test "p ((int (*) ()) $f) ()" " = 0"
>   
> -# Call the function, casting the function result to the correct type.
> -gdb_test "p (int) foo ()" " = 0"
> +    # Call the function, casting the function result to the correct type.
> +    gdb_test "p (int) $f ()" " = 0"
> +}
> 
> base-commit: 73b22419edd563cf26bc8688c3f04fb3cf414cb0


  reply	other threads:[~2023-10-16  9:21 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-28 16:07 [PATCH] " Tom de Vries
2023-10-16  9:21 ` Tom de Vries [this message]
2023-10-16 13:55 ` Andrew Burgess

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=786e514c-6ff7-452e-a52e-5d727942d1c0@suse.de \
    --to=tdevries@suse.de \
    --cc=gdb-patches@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).