SYMBOL_LANGUAGE assertion failure on AIX

SYMBOL_LANGUAGE assertion failure on AIX

[Sangamesh Mallayya]

[-- Attachment #1: Type: text/plain, Size: 2292 bytes --]

Hi All,

Lot of gdb testcases are failing due assertion failures on AIX with the 
error below. 

(gdb) br main
Breakpoint 1 at 0x1000075c: file 
/home/gdb_build/binutils-gdb/gdb/testsuite/gdb.base/savedregs.c, line 67.
(gdb) r
Starting program: 
/home/gdb_build/binutils-gdb/gdb/testsuite/outputs/gdb.base/savedregs/savedregs 


Breakpoint 1, main ()
    at /home/gdb_build/binutils-gdb/gdb/testsuite/gdb.base/savedregs.c:67
67        signal (SIGILL, catcher);
(gdb) s
dictionary.c:690: internal-error: void insert_symbol_hashed(dictionary*, 
symbol*): Assertion `SYMBOL_LANGUAGE (sym) == DICT_LANGUAGE 
(dict)->la_language' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) y


This error seems to be happening after the fix 
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=5ffa0793690b42b2a0c1c21dbb5e64634e58fa00


Attached patch resolves the assertion failure and  gdb is able to debug 
the program further.

(gdb) br main
Breakpoint 1 at 0x1000075c: file 
/home/gdb_build/binutils-gdb/gdb/testsuite/gdb.base/savedregs.c, line 67.
(gdb) r
Starting program: 
/home/gdb_build/binutils-gdb/gdb/testsuite/outputs/gdb.base/savedregs/savedregs 


Breakpoint 1, main ()
    at /home/gdb_build/binutils-gdb/gdb/testsuite/gdb.base/savedregs.c:67
67        signal (SIGILL, catcher);
(gdb) s
warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

68        signal (SIGSEGV, catcher);
(gdb) s
warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

warning: (Internal error: pc 0x10000804 in read in psymtab, but not in 
symtab.)

69        thrower ();
(gdb) s
thrower ()
    at /home/gdb_build/binutils-gdb/gdb/testsuite/gdb.base/savedregs.c:49
49        *(char *)0 = 0;
(gdb) 



Please let me know the comments.



Thanks,
Sangamesh

[-- Attachment #2: aix_assertion_failure.patch --]
[-- Type: application/octet-stream, Size: 1582 bytes --]

--- ./gdb/xcoffread.c_orig	2018-10-15 04:03:45 +0000
+++ ./gdb/xcoffread.c	2018-10-15 06:44:59 +0000
@@ -1022,6 +1022,7 @@
   int just_started = 1;
   int depth = 0;
   CORE_ADDR fcn_start_addr = 0;
+  enum language pst_symtab_language;
 
   struct coff_symbol fcn_stab_saved = { 0 };
 
@@ -1042,10 +1043,11 @@
 
   set_last_source_file (NULL);
   last_csect_name = 0;
+  pst_symtab_language = deduce_language_from_filename(filestring);
 
   start_stabs ();
   start_symtab (objfile, filestring, (char *) NULL, file_start_addr,
-		language_unknown);
+		pst_symtab_language);
   record_debugformat (debugfmt);
   symnum = ((struct symloc *) pst->read_symtab_private)->first_symnum;
   max_symnum =
@@ -1138,8 +1140,8 @@
 	    }
 
 	  start_stabs ();
-	  start_symtab (objfile, "_globals_", (char *) NULL, (CORE_ADDR) 0,
-			language_unknown);
+	  start_symtab (objfile, "_globals_", (char *) NULL,
+			(CORE_ADDR) 0, pst_symtab_language);
 	  record_debugformat (debugfmt);
 	  cur_src_end_addr = first_object_file_end;
 	  /* Done with all files, everything from here on is globals.  */
@@ -1229,7 +1231,7 @@
 			  /* Give all csects for this source file the same
 			     name.  */
 			  start_symtab (objfile, filestring, NULL,
-					(CORE_ADDR) 0, language_unknown);
+					(CORE_ADDR) 0, pst_symtab_language);
 			  record_debugformat (debugfmt);
 			}
 
@@ -1350,7 +1352,7 @@
 
 	  start_stabs ();
 	  start_symtab (objfile, filestring, (char *) NULL, (CORE_ADDR) 0,
-			language_unknown);
+			pst_symtab_language);
 	  record_debugformat (debugfmt);
 	  last_csect_name = 0;
 

Re: SYMBOL_LANGUAGE assertion failure on AIX

[Ulrich Weigand]

Sangamesh Mallaya wrote:

> Lot of gdb testcases are failing due assertion failures on AIX with the
> error below.

The patch makes sense to me.

You'll need a ChangeLog entry; patch is OK with that.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU/Linux compilers and toolchain
  Ulrich.Weigand@de.ibm.com

Re: [PATCH] SYMBOL_LANGUAGE assertion failure on AIX

[Sangamesh Mallayya]

[-- Attachment #1: Type: text/plain, Size: 732 bytes --]

Hi Ulrich

Thanks for the review and comments.
Attaching the changelog file and git formatted patch.

 

Thanks,
Sangamesh



From:   "Ulrich Weigand" <uweigand@de.ibm.com>
To:     Sangamesh Mallayya/India/IBM@IBMIN
Cc:     gdb-patches@sourceware.org, Ulrich.Weigand@de.ibm.com (Ulrich 
Weigand)
Date:   10/18/2018 08:45 PM
Subject:        Re: SYMBOL_LANGUAGE assertion failure on AIX



Sangamesh Mallaya wrote:

> Lot of gdb testcases are failing due assertion failures on AIX with the
> error below.

The patch makes sense to me.

You'll need a ChangeLog entry; patch is OK with that.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU/Linux compilers and toolchain
  Ulrich.Weigand@de.ibm.com



[-- Attachment #2: aix_assertion_failure.patch --]
[-- Type: application/octet-stream, Size: 1963 bytes --]

diff --git a/gdb/xcoffread.c b/gdb/xcoffread.c
index 941471a..acf0959 100644
--- a/gdb/xcoffread.c
+++ b/gdb/xcoffread.c
@@ -1022,6 +1022,7 @@ read_xcoff_symtab (struct objfile *objfile, struct partial_symtab *pst)
   int just_started = 1;
   int depth = 0;
   CORE_ADDR fcn_start_addr = 0;
+  enum language pst_symtab_language;
 
   struct coff_symbol fcn_stab_saved = { 0 };
 
@@ -1042,10 +1043,11 @@ read_xcoff_symtab (struct objfile *objfile, struct partial_symtab *pst)
 
   set_last_source_file (NULL);
   last_csect_name = 0;
+  pst_symtab_language = deduce_language_from_filename (filestring);
 
   start_stabs ();
   start_symtab (objfile, filestring, (char *) NULL, file_start_addr,
-		language_unknown);
+		pst_symtab_language);
   record_debugformat (debugfmt);
   symnum = ((struct symloc *) pst->read_symtab_private)->first_symnum;
   max_symnum =
@@ -1138,8 +1140,8 @@ read_xcoff_symtab (struct objfile *objfile, struct partial_symtab *pst)
 	    }
 
 	  start_stabs ();
-	  start_symtab (objfile, "_globals_", (char *) NULL, (CORE_ADDR) 0,
-			language_unknown);
+	  start_symtab (objfile, "_globals_", (char *) NULL,
+			(CORE_ADDR) 0, pst_symtab_language);
 	  record_debugformat (debugfmt);
 	  cur_src_end_addr = first_object_file_end;
 	  /* Done with all files, everything from here on is globals.  */
@@ -1229,7 +1231,7 @@ read_xcoff_symtab (struct objfile *objfile, struct partial_symtab *pst)
 			  /* Give all csects for this source file the same
 			     name.  */
 			  start_symtab (objfile, filestring, NULL,
-					(CORE_ADDR) 0, language_unknown);
+					(CORE_ADDR) 0, pst_symtab_language);
 			  record_debugformat (debugfmt);
 			}
 
@@ -1350,7 +1352,7 @@ read_xcoff_symtab (struct objfile *objfile, struct partial_symtab *pst)
 
 	  start_stabs ();
 	  start_symtab (objfile, filestring, (char *) NULL, (CORE_ADDR) 0,
-			language_unknown);
+			pst_symtab_language);
 	  record_debugformat (debugfmt);
 	  last_csect_name = 0;
 

[-- Attachment #3: ChangeLog --]
[-- Type: application/octet-stream, Size: 180 bytes --]

2018-10-23 Sangamesh Mallayya <sangamesh.swamy@in.ibm.com>
	    Ulrich Weigand  <uweigand@de.ibm.com>
	
	* xcoffread.c (read_xcoff_symtab): Fix SYMBOL_LANGUAGE assertion
	failure.

Re: [PATCH] SYMBOL_LANGUAGE assertion failure on AIX

[Ulrich Weigand]

Sangamesh Mallaya wrote:

> Thanks for the review and comments.
> Attaching the changelog file and git formatted patch.

>	* xcoffread.c (read_xcoff_symtab): Fix SYMBOL_LANGUAGE assertion
>	failure.

GNU coding style is to mention in the ChangeLog *what* was changed,
not why the change is done.  "Fix assertion failure" doesn't describe
*what* actually was changed, so it should read more like this:

	* xcoffread.c (read_xcoff_symtab): Pass deduced language to
	symtab_start instead of always using language_unknown.

(You can and should still describe the reason for the change, e.g. that
it fixes an assertion failure, in the git commit message.)

You should have git commit access yourself now, right?  Feel free to
commit the change now.  (If you don't have access, please let me know.)

Thanks,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU/Linux compilers and toolchain
  Ulrich.Weigand@de.ibm.com

Re: [PATCH] SYMBOL_LANGUAGE assertion failure on AIX

[Sangamesh Mallayya]

Hi Ulrich,

Thanks for the help.
I am able to commit the changes 
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=b971899198607b844f5a37e39dc561766c3b331a

Thanks,
Sangamesh



From:   "Ulrich Weigand" <uweigand@de.ibm.com>
To:     Sangamesh Mallayya/India/IBM@IBMIN
Cc:     keiths@redhat.com (Keith Seitz), gdb-patches@sourceware.org
Date:   10/24/2018 11:34 PM
Subject:        Re: [PATCH] SYMBOL_LANGUAGE assertion failure on AIX



Sangamesh Mallaya wrote:

> Thanks for the review and comments.
> Attaching the changelog file and git formatted patch.

>                * xcoffread.c (read_xcoff_symtab): Fix SYMBOL_LANGUAGE 
assertion
>                failure.

GNU coding style is to mention in the ChangeLog *what* was changed,
not why the change is done.  "Fix assertion failure" doesn't describe
*what* actually was changed, so it should read more like this:

                 * xcoffread.c (read_xcoff_symtab): Pass deduced language 
to
                 symtab_start instead of always using language_unknown.

(You can and should still describe the reason for the change, e.g. that
it fixes an assertion failure, in the git commit message.)

You should have git commit access yourself now, right?  Feel free to
commit the change now.  (If you don't have access, please let me know.)

Thanks,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU/Linux compilers and toolchain
  Ulrich.Weigand@de.ibm.com