[PATCH v2 00/42] Share DWARF partial symtabs between objfiles
Simon Marchi
simark@simark.ca
Wed May 13 15:46:48 GMT 2020
On 2020-05-13 6:17 a.m., Tom de Vries wrote:
> On 12-05-2020 23:08, Simon Marchi via Gdb-patches wrote:
>> This is version 2 of this patchset originally written by Tom:
>>
>> https://sourceware.org/legacy-ml/gdb-patches/2020-02/msg00594.html
>>
>> The current patchset therefore contains a mix of Tom's patches that I modified,
>> plus a few new patches. I won't repeat everything that is said in the cover
>> letter of that patchset, please read that to understand the motivation behind
>> this change..
>>
>> The big difference between v1 and v2 is how the split is done. In v1, the
>> object structure looks like this:
>>
>> objfile -> dwarf2_per_objfile -> dwarf2_unshareable
>>
>> where `dwarf2_per_objfile` is actually shared between objfiles (there is one
>> copy per BFD), and `dwarf2_unshareable` is not (there is one copy per objfile).
>> The dwarf2_per_objfile -> dwarf2_unshareable link is updated at all entry
>> points of the DWARF code based on what is the current objfile.
>>
>> The current patchset (v2) proposes this structure instead, which in my opinion
>> is more natural:
>>
>> objfile -> dwarf2_per_objfile -> dwarf2_per_bfd
>>
>> Where `dwarf2_per_objfile` is really per-objfile (there is one copy per
>> objfile) and `dwarf2_per_bfd` is per-bfd (there is one copy per-bfd). There is
>> no link to update, whatever is shared between objfiles is put in
>> `dwarf2_per_bfd` and each `dwarf2_per_objfile` always points to one
>> `dwarf2_per_bfd`.
>>
>> Since there were performance numbers with the original patchset, I made my own
>> here to see if the speedups were still there. Like in the original patchset, I
>> set "maintenance time 1" and did "add-inferior -exec ./gdb" after "./gdb gdb".
>> I disabled any auto-loading to avoid loading gdb-gdb.{gdb,py}. GDB is built
>> with -O2 and gcc 9, and it is running on an AMD EPYC 7351P.
>>
>> Before: Command execution time: 1.213480 (cpu), 1.214268 (wall)
>> After: Command execution time: 0.001697 (cpu), 0.001694 (wall)
>>
>> These are the times of one run, but all runs I did are very close. The "after"
>> time is surpringly low compared to the original numbers, I hope it doesn't mean
>> I messed up something and we are skipping something important...
>>
>> At some point I tested with all these boards and did not see any regression
>> (they helped very much to find some issues in the process though):
>>
>> unix cc-with-debug-names cc-with-dwz cc-with-dwz-m cc-with-gdb-index dwarf4-gdb-index fission-dwp fission readnow
>>
>> I have since rebased the patchset and resolved conflicts, it's therefore
>> possible I messed something up.
>>
>
> Hi, I applied the patch series on top of commit 90d00bbd9c "Sync config
> and libiberty with GCC", and ran into the following:
> ...
> (gdb) PASS: gdb.ada/variant.exp: scenario=minimal: print nav1
> print nav2^M
> ERROR: GDB process no longer exists
> GDB process exited with wait status 8901 exp9 0 0 CHILDKILLED SIGABRT
> SIGABRT
> UNRESOLVED: gdb.ada/variant.exp: scenario=minimal: print nav2
> ...
>
> Reproduced on command line using:
> ...
> $ gdb -batch ./outputs/gdb.ada/variant/pkg \
> -ex 'break pkg.adb:40' \
> -ex run \
> -ex 'print nav2'
> Breakpoint 1 at 0x403047: file gdb/testsuite/gdb.ada/variant/pkg.adb,
> line 40.
>
> Breakpoint 1, pkg () at gdb/testsuite/gdb.ada/variant/pkg.adb:40
> 40 null; -- STOP
> Aborted (core dumped)
> ...
>
> Using gdb, we get more info:
> ...
> Thread 1 "gdb" received signal SIGSEGV, Segmentation fault.
> 0x00000000005b9d02 in dwarf2_locexpr_baton_eval (dlbaton=0x7fffffffc608,
> frame=0x15c75f0,
> addr_stack=0x7fffffffc790, valp=0x7fffffffc638, push_initial_value=true)
> at /data/gdb_versions/devel/src/gdb/dwarf2/loc.c:2490
> 2490 ctx.gdbarch = per_objfile->objfile->arch ();
> ...
>
> The problem seems to be:
> ...
> (gdb) p per_objfile
> $1 = (dwarf2_per_objfile *) 0xffffffffffffffff
> ...
>
> Backtrace:
> ...
> (gdb) bt
> #0 0x00000000005b9d02 in dwarf2_locexpr_baton_eval
> (dlbaton=0x7fffffffc608, frame=0x15c75f0,
> addr_stack=0x7fffffffc790, valp=0x7fffffffc638, push_initial_value=true)
> at /data/gdb_versions/devel/src/gdb/dwarf2/loc.c:2490
> #1 0x00000000005ba007 in dwarf2_evaluate_property (prop=0x7fffffffc640,
> frame=0x15c75f0,
> addr_stack=0x7fffffffc790, value=0x7fffffffc638,
> push_initial_value=true)
> at /data/gdb_versions/devel/src/gdb/dwarf2/loc.c:2562
> #2 0x0000000000672aff in resolve_dynamic_struct (type=0x1b25660,
> addr_stack=0x7fffffffc790)
> at /data/gdb_versions/devel/src/gdb/gdbtypes.c:2480
> #3 0x000000000067309c in resolve_dynamic_type_internal (type=0x1b25660,
> addr_stack=0x7fffffffc790, top_level=1)
> at /data/gdb_versions/devel/src/gdb/gdbtypes.c:2613
> #4 0x0000000000673225 in resolve_dynamic_type (type=0x1b25660,
> valaddr=..., addr=4255376)
> at /data/gdb_versions/devel/src/gdb/gdbtypes.c:2649
> #5 0x00000000009b1918 in value_from_contents_and_address
> (type=0x1b25660, valaddr=0x0, address=4255376)
> at /data/gdb_versions/devel/src/gdb/value.c:3483
> #6 0x0000000000999f9c in get_value_at (type=0x1b25660, addr=4255376,
> lazy=1)
> at /data/gdb_versions/devel/src/gdb/valops.c:899
> #7 0x000000000099a00a in value_at_lazy (type=0x1b25660, addr=4255376)
> at /data/gdb_versions/devel/src/gdb/valops.c:936
> #8 0x000000000064dc01 in default_read_var_value (var=0x1b256c0,
> var_block=0x1b25760, frame=0x0)
> at /data/gdb_versions/devel/src/gdb/findvar.c:800
> #9 0x000000000043a2f2 in ada_read_var_value (var=0x1b256c0,
> var_block=0x1b25760, frame=0x0)
> at /data/gdb_versions/devel/src/gdb/ada-lang.c:14061
> #10 0x000000000064dccd in read_var_value (var=0x1b256c0,
> var_block=0x1b25760, frame=0x0)
> at /data/gdb_versions/devel/src/gdb/findvar.c:815
> #11 0x000000000099acf3 in value_of_variable (var=0x1b256c0, b=0x1b25760)
> --Type <RET> for more, q to quit, c to continue without paging--
> at /data/gdb_versions/devel/src/gdb/valops.c:1293
> #12 0x000000000062ddd8 in evaluate_var_value
> (noside=EVAL_AVOID_SIDE_EFFECTS, blk=0x1b25760, var=0x1b256c0)
> at /data/gdb_versions/devel/src/gdb/eval.c:716
> #13 0x000000000062f4e3 in evaluate_subexp_standard (expect_type=0x0,
> exp=0x1bcf8c0, pos=0x7fffffffd1c4,
> noside=EVAL_AVOID_SIDE_EFFECTS) at
> /data/gdb_versions/devel/src/gdb/eval.c:1319
> #14 0x000000000043232e in ada_evaluate_subexp (expect_type=0x0,
> exp=0x1bcf8c0, pos=0x7fffffffd1c4,
> noside=EVAL_AVOID_SIDE_EFFECTS) at
> /data/gdb_versions/devel/src/gdb/ada-lang.c:10656
> #15 0x000000000062ca12 in evaluate_subexp (expect_type=0x0,
> exp=0x1bcf8c0, pos=0x7fffffffd1c4,
> noside=EVAL_AVOID_SIDE_EFFECTS) at
> /data/gdb_versions/devel/src/gdb/eval.c:77
> #16 0x000000000042f787 in evaluate_subexp_type (exp=0x1bcf8c0,
> pos=0x7fffffffd1c4)
> at /data/gdb_versions/devel/src/gdb/ada-lang.c:9417
> #17 0x0000000000423e80 in resolve_subexp (expp=0x7fffffffd418,
> pos=0x7fffffffd1c4, deprocedure_p=1,
> context_type=0x0, parse_completion=0, tracker=0x7fffffffd2f0)
> at /data/gdb_versions/devel/src/gdb/ada-lang.c:3826
> #18 0x0000000000422fd2 in resolve (expp=0x7fffffffd418,
> void_context_p=0, parse_completion=0,
> tracker=0x7fffffffd2f0) at
> /data/gdb_versions/devel/src/gdb/ada-lang.c:3463
> #19 0x00000000007bd06c in parse_exp_in_context
> (stringptr=0x7fffffffd3d0, pc=0, block=0x0, comma=0,
> void_context_p=0, out_subexp=0x0, tracker=0x7fffffffd2f0, cstate=0x0)
> at /data/gdb_versions/devel/src/gdb/parse.c:1149
> #20 0x00000000007bcd28 in parse_exp_1 (stringptr=0x7fffffffd3d0, pc=0,
> block=0x0, comma=0, tracker=0x0)
> at /data/gdb_versions/devel/src/gdb/parse.c:1031
> #21 0x00000000007bd1cf in parse_expression (string=0x7fffffffe0fb
> "nav2", tracker=0x0)
> at /data/gdb_versions/devel/src/gdb/parse.c:1167
> #22 0x00000000007c04b6 in print_command_1 (args=0x7fffffffe0fb "nav2",
> voidprint=1)
> at /data/gdb_versions/devel/src/gdb/printcmd.c:1214
> --Type <RET> for more, q to quit, c to continue without paging--
> #23 0x00000000007c062c in print_command (exp=0x7fffffffe0fb "nav2",
> from_tty=0)
> at /data/gdb_versions/devel/src/gdb/printcmd.c:1244
> #24 0x00000000004fdf16 in do_const_cfunc (c=0x19cca60,
> args=0x7fffffffe0fb "nav2", from_tty=0)
> at /data/gdb_versions/devel/src/gdb/cli/cli-decode.c:107
> #25 0x000000000050103f in cmd_func (cmd=0x19cca60, args=0x7fffffffe0fb
> "nav2", from_tty=0)
> at /data/gdb_versions/devel/src/gdb/cli/cli-decode.c:2004
> #26 0x000000000091bcf5 in execute_command (p=0x7fffffffe0fe "2", from_tty=0)
> at /data/gdb_versions/devel/src/gdb/top.c:655
> #27 0x0000000000746633 in catch_command_errors (command=0x91b889
> <execute_command(char const*, int)>,
> arg=0x7fffffffe0f5 "print nav2", from_tty=0) at
> /data/gdb_versions/devel/src/gdb/main.c:457
> #28 0x0000000000747a1b in captured_main_1 (context=0x7fffffffd9e0)
> at /data/gdb_versions/devel/src/gdb/main.c:1219
> #29 0x0000000000747c10 in captured_main (data=0x7fffffffd9e0) at
> /data/gdb_versions/devel/src/gdb/main.c:1244
> #30 0x0000000000747c7b in gdb_main (args=0x7fffffffd9e0) at
> /data/gdb_versions/devel/src/gdb/main.c:1269
> #31 0x000000000041520e in main (argc=14, argv=0x7fffffffdae8) at
> /data/gdb_versions/devel/src/gdb/gdb.c:32
Thanks for the report. We were missing assigning dlbaton->per_objfile in
handle_data_member_location. This is some recent code on which I rebased
just before sending the series. Good thing I included that disclaimer :).
I fixed it in patch 4, "Add dwarf2_per_objfile member to DWARF batons".
Simon
More information about the Gdb-patches
mailing list