Bug 30261

After clean rebuild, reproduces as a hang instead.

Now, again as a segfault, but different location:
...
(gdb) maintenance wait-for-index-cache^M
0x595ac6 gdb_internal_backtrace_1^M
        /data/vries/gdb/src/gdb/bt-utils.c:122^M
  ...
0x41feea _ZNK4type4codeEv^M
        /data/vries/gdb/src/gdb/gdbtypes.h:912^M
...

Hmm, the test already doesn't work with "maint set worker-threads 0" on native.

(In reply to Tom de Vries from comment #3)
> Hmm, the test already doesn't work with "maint set worker-threads 0" on
> native.

Likewise in test-case gdb.dwarf2/per-bfd-sharing.exp, there we run into:
...
FAIL: gdb.dwarf2/per-bfd-sharing.exp: couldn't remove files in temporary cache dir
...

At commit f0c3dcc1ca7 ("Ensure index cache entry written in test"):
...
$ for n in $(seq 1 10); do ./test.sh -native-extended-gdbserver | grep '^#' | sort -u; done
# of expected passes            23
# of unresolved testcases       1
# of expected passes            24
# of expected passes            24
# of expected passes            24
# of expected passes            24
# of expected passes            22
# of unexpected core files      1
# of unresolved testcases       2
# of expected passes            24
# of expected passes            24
# of expected passes            24
# of expected passes            24
...

At commit 52e5e48e533 ("Write the DWARF index in the background"):
...
$ for n in $(seq 1 10); do ./test.sh -native-extended-gdbserver | grep '^#' | sort -u; done
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            17
# of unexpected core files      1
# of unexpected failures        3
# of expected passes            17
# of unexpected core files      1
# of unexpected failures        2
# of unresolved testcases       1
# of expected passes            16
# of unexpected core files      1
# of unexpected failures        4
# of expected passes            19
# of expected passes            17
# of unexpected core files      1
# of unresolved testcases       2
# of expected passes            17
# of unexpected core files      1
# of unexpected failures        3
# of expected passes            19
...

At commit 52e5e48e533^:
...
$ for n in $(seq 1 10); do ./test.sh -native-extended-gdbserver | grep '^#' | sort -u; done
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
# of expected passes            19
...

So, this looks like a regression by commit 52e5e48e533 ("Write the DWARF index in the background") that hasn't been fixed by commit f0c3dcc1ca7 ("Ensure index cache entry written in test").

Created attachment 14775 [details]
gdb.log with thread sanitizer

$ grep ThreadS gdb.log 
WARNING: ThreadSanitizer: data race (pid=22270)
SUMMARY: ThreadSanitizer: data race /data/vries/gdb/src/gdb/dwarf2/index-cache.c:76 in index_cache::enable()
WARNING: ThreadSanitizer: data race (pid=22359)
SUMMARY: ThreadSanitizer: data race /usr/include/c++/7/bits/unique_ptr.h:147 in std::__uniq_ptr_impl<dwarf_scanner_base, std::default_delete<dwarf_scanner_base> >::_M_ptr() const
WARNING: ThreadSanitizer: data race (pid=22359)
SUMMARY: ThreadSanitizer: data race /data/vries/gdb/src/gdb/dwarf2/index-write.c:1229 in write_gdbindex
WARNING: ThreadSanitizer: data race (pid=22404)
SUMMARY: ThreadSanitizer: data race /data/vries/gdb/src/bfd/cache.c:585 in bfd_open_file
WARNING: ThreadSanitizer: data race (pid=22404)
SUMMARY: ThreadSanitizer: data race /usr/include/c++/7/bits/unique_ptr.h:147 in std::__uniq_ptr_impl<dwarf_scanner_base, std::default_delete<dwarf_scanner_base> >::_M_ptr() const
WARNING: ThreadSanitizer: data race (pid=22404)
SUMMARY: ThreadSanitizer: data race /data/vries/gdb/src/gdb/dwarf2/index-write.c:1229 in write_gdbindex
...

I'll investigate.

Ok, I found the problem.
This would also be fixed by my WIP to move more reading to the background,
but unfortunately that series isn't quite ready yet.

https://sourceware.org/pipermail/gdb-patches/2023-March/198290.html

The master branch has been updated by Tom Tromey <tromey@sourceware.org>:

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6f214d0f399847b13f979651c3b46befcbb42140

commit 6f214d0f399847b13f979651c3b46befcbb42140
Author: Tom Tromey <tom@tromey.com>
Date:   Fri Mar 24 15:53:22 2023 -0600

    Fix race in background index-cache writing
    
    Tom de Vries pointed out a bug in the index-cache background writer --
    sometimes it will fail.  He also noted that it fails when the number
    of worker threads is set to zero.  These turn out to be the same
    problem -- the cache can't be written to until the per-BFD's
    "index_table" member is set.
    
    This patch avoids the race by rearranging the code slightly, to ensure
    the cache cannot possibly be written before the member is set.
    
    Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30261

Fixed.