Re: [PATCH v3 1/2] gdb/record: print frame information when exiting a recursive call

[Kevin Buettner <kevinb@redhat.com>]

Hi Gwen,

On Wed, 27 Sep 2023 12:19:51 +0200
Guinevere Larsen <blarsen@redhat.com> wrote:

>  gdb/infrun.c                            | 18 ++++++++++
>  gdb/testsuite/gdb.reverse/recursion.c   | 44 ++++++++++++++++++++++++
>  gdb/testsuite/gdb.reverse/recursion.exp | 45 +++++++++++++++++++++++++
>  3 files changed, 107 insertions(+)
>  create mode 100644 gdb/testsuite/gdb.reverse/recursion.c
>  create mode 100644 gdb/testsuite/gdb.reverse/recursion.exp

I've applied your patch and have tested on Fedora 38, for both
x86_64 and aarch64 (both native).  I see no regressions on aarch64,
but on x86_64, I'm seeing this failure when your patch is applied:

FAIL: gdb.dwarf2/dw2-out-of-range-end-of-seq.exp: END with address 1 eliminated

After studying the log files, this makes absolutely no sense to me
since your patch should not have any affect on the "maint info line-table"
command.  Here are the relevant sections of the log files:

- - - passing run - - -
(gdb) break -qualified main
Breakpoint 1 at 0x40110a: file /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/main.c, line 1.
(gdb) run 
Starting program: /mesquite2/sourceware-git/f38-review/bld/gdb/testsuite/outputs/gdb.dwarf2/dw2-out-of-range-end-of-seq/dw2-out-of-range-end-of-seq 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".

Breakpoint 1, 0x000000000040110a in main () at /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/main.c:1
1	/* This testcase is part of GDB, the GNU debugger.
(gdb) maint info line-table main.c$
objfile: /mesquite2/sourceware-git/f38-review/bld/gdb/testsuite/outputs/gdb.dwarf2/dw2-out-of-range-end-of-seq/dw2-out-of-range-end-of-seq ((struct objfile *) 0x18e5670)
compunit_symtab: main.c ((struct compunit_symtab *) 0x18e5b90)
symtab: /mesquite2/sourceware-git/worktree-review/gdb/testsuite/gdb.dwarf2/main.c ((struct symtab *) 0x18e5c10)
linetable: ((struct linetable *) 0x18e5e30):
INDEX  LINE   REL-ADDRESS        UNREL-ADDRESS      IS-STMT PROLOGUE-END 
0      1      0x0000000000401106 0x0000000000401106 Y                    
1      END    0x0000000000401111 0x0000000000401111 Y                    

(gdb) PASS: gdb.dwarf2/dw2-out-of-range-end-of-seq.exp: END with address 1 eliminated
testcase /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/dw2-out-of-range-end-of-seq.exp completed in 1 seconds
- - - end passing run - - -

- - - failing run - - -
(gdb) break -qualified main
Breakpoint 1 at 0x40110a: file /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/main.c, line 1.
(gdb) run 
Starting program: /mesquite2/sourceware-git/f38-review/bld/gdb/testsuite/outputs/gdb.dwarf2/dw2-out-of-range-end-of-seq/dw2-out-of-range-end-of-seq 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".

Breakpoint 1, 0x000000000040110a in main () at /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/main.c:1
1	/* This testcase is part of GDB, the GNU debugger.
(gdb) maint info line-table main.c$
objfile: /mesquite2/sourceware-git/f38-review/bld/gdb/testsuite/outputs/gdb.dwarf2/dw2-out-of-range-end-of-seq/dw2-out-of-range-end-of-seq ((struct objfile *) 0x24a85b0)
compunit_symtab: main.c ((struct compunit_symtab *) 0x24a8ad0)
symtab: /mesquite2/sourceware-git/worktree-review/gdb/testsuite/gdb.dwarf2/main.c ((struct symtab *) 0x24a8b50)
linetable: ((struct linetable *) 0x24a8d70):
INDEX  LINE   REL-ADDRESS        UNREL-ADDRESS      IS-STMT PROLOGUE-END 
0      1      0x0000000000401106 0x0000000000401106 Y                    
1      END    0x0000000000401111 0x0000000000401111 Y                    

objfile: /usr/lib/debug/usr/lib64/libc.so.6-2.37-4.fc38.x86_64.debug ((struct objfile *) 0x26435b0)
compunit_symtab: <unknown> ((struct compunit_symtab *) 0x2eb2260)
symtab: /usr/src/debug/glibc-2.37-4.fc38.x86_64/intl/textdomain.c ((struct symtab *) 0x2f60430)
linetable: ((struct linetable *) 0x0):
No line table.

objfile: /usr/lib/debug/usr/lib64/libc.so.6-2.37-4.fc38.x86_64.debug ((struct objfile *) 0x26435b0)
compunit_symtab: <unknown> ((struct compunit_symtab *) 0x2eb2260)
symtab: /usr/src/debug/glibc-2.37-4.fc38.x86_64/support/support_test_main.c ((struct symtab *) 0x2f725d0)
linetable: ((struct linetable *) 0x0):
No line table.

(gdb) FAIL: gdb.dwarf2/dw2-out-of-range-end-of-seq.exp: END with address 1 eliminated
testcase /ironwood1/sourceware-git/f38-review/bld/../../worktree-review/gdb/testsuite/gdb.dwarf2/dw2-out-of-range-end-of-seq.exp completed in 1 seconds
- - - end failing run - - -

The difference here is that the failing run is (also) attempting to
list the line table for libc (and twice for some reason).  As noted
earlier, I don't see how your change can affect this area of the code,
but, multiple times, I've done a "git reset --hard" to get rid of your
commit, followed by a rebuild and test, each time seeing the test pass. 
After that, I've applied your patch, rebuilt, and tested again, each
time seeing this test fail.

Are you able to reproduce this problem?

(I'll continue to study it here - there may be something weird going
on with the VM upon which I'm testing...)

Kevin