gdb/ctf: don't create a buildsym_compunit when building partial symbols
authorSimon Marchi <simon.marchi@polymtl.ca>
Thu, 31 Mar 2022 21:20:17 +0000 (17:20 -0400)
committerSimon Marchi <simon.marchi@polymtl.ca>
Fri, 1 Apr 2022 01:26:48 +0000 (21:26 -0400)
commit8458fb44b45ed25ae08237771fa4ada9a8f463f4
tree66c76aa8cbabc70823124e72e884c8747eede9c5
parent8839e3f3b0d1af5d93ba1d4f667ce33936bb7db2
gdb/ctf: don't create a buildsym_compunit when building partial symbols

I am trying to do some changes to buildsym_compunit, so I am auditing
the current uses.  Something seems odd with this use of
buildsym_compunit (that this patch removes).

A buildsym_compunit is normally used when building a compunit_symtab.
That is, when expanding a partial symtab into a full compunit symtab.
In ctfread.c, a buildsym_compunit is created in ctf_start_archive, which
is only used when creating partial symtabs.  At this moment, I don't
see how that's useful.  ctf_start_archive creates a new
buildsym_compunit and starts a subfile.  But that buildsym_compunit is
never used again.  It's just overriden in ctf_start_symtab, which means
we leak the old buildsym_compunit, I suppose.

Remove ctf_start_archive completely.  Add an assert in
ctf_start_symtab to verify that we are not overwriting an existing
buildsym_compunit (meaning we'd leak the existing one).  This assert
triggers without the other part of the fix.  When doing:

  $ ./gdb --data-directory=data-directory /tmp/babeltrace-ctf/src/lib/.libs/libbabeltrace2.so.0.0.0
  ...
  (gdb) maintenance expand-symtabs
  /home/simark/src/binutils-gdb/gdb/ctfread.c:1255: internal-error: ctf_start_symtab: Assertion `!ccp->builder' failed.

Change-Id: I666d146454a019f08e7305f3a1c4a974d27b4592
gdb/ctfread.c