mbox series

[0/7] Changes to gdb-index creation

Message ID cover.1701107594.git.aburgess@redhat.com
Headers
Series Changes to gdb-index creation |

Message

Andrew Burgess Nov. 27, 2023, 5:55 p.m. UTC 
  Patches #1 and #2 are pretty straight forward improvements; don't
think they will be controversial.

Patch #4 is an interesting bug fix.

Patches #6 and #7 are what this series is really all about.  Making
the indexes created by 'save gdb-index' be determenistic as the number
of worker-threads that GDB uses changes.


---

Andrew Burgess (7):
  gdb: allow use of ~ in 'save gdb-index' command
  gdb: option completion for 'save gdb-index' command
  gdb/testsuite: small refactor in selftest-support.exp
  gdb: reduce size of generated gdb-index file
  gdb: C++-ify mapped_symtab from dwarf2/index-write.c
  gdb: generate gdb-index identically regardless of work thread count
  gdb: generate dwarf-5 index identically as worker-thread count changes

 gdb/dwarf2/index-write.c                     | 315 ++++++++++++++-----
 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp |  91 ++++++
 gdb/testsuite/gdb.gdb/index-file.exp         | 162 ++++++++++
 gdb/testsuite/lib/gdb.exp                    |  15 +
 gdb/testsuite/lib/selftest-support.exp       |  32 +-
 5 files changed, 536 insertions(+), 79 deletions(-)
 create mode 100644 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp
 create mode 100644 gdb/testsuite/gdb.gdb/index-file.exp


base-commit: 935dc9ff652ca256c10672412c1df3da95cadbfb

Comments

Tom Tromey Nov. 27, 2023, 7:38 p.m. UTC | #1 
>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

Andrew> Patches #1 and #2 are pretty straight forward improvements; don't
Andrew> think they will be controversial.

Andrew> Patch #4 is an interesting bug fix.

Andrew> Patches #6 and #7 are what this series is really all about.  Making
Andrew> the indexes created by 'save gdb-index' be determenistic as the number
Andrew> of worker-threads that GDB uses changes.

I read through this.  I was a bit worried about new gdb.gdb tests (I
feel those are often more trouble than they're worth), but this one
avoids the problem stuff and anyway there doesn't seem to be a better
way to do it.

Anyway, I sent a nit or two, everything looks good to me.
Thanks for doing this, and nice find on the hash table bug.

Approved-By: Tom Tromey <tom@tromey.com>

Tom
Andrew Burgess Nov. 28, 2023, 10:34 a.m. UTC | #2 
Tom Tromey <tom@tromey.com> writes:

>>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:
>
> Andrew> Patches #1 and #2 are pretty straight forward improvements; don't
> Andrew> think they will be controversial.
>
> Andrew> Patch #4 is an interesting bug fix.
>
> Andrew> Patches #6 and #7 are what this series is really all about.  Making
> Andrew> the indexes created by 'save gdb-index' be determenistic as the number
> Andrew> of worker-threads that GDB uses changes.
>
> I read through this.  I was a bit worried about new gdb.gdb tests (I
> feel those are often more trouble than they're worth), but this one
> avoids the problem stuff and anyway there doesn't seem to be a better
> way to do it.
>
> Anyway, I sent a nit or two, everything looks good to me.
> Thanks for doing this, and nice find on the hash table bug.
>
> Approved-By: Tom Tromey <tom@tromey.com>

I pushed this series.

Thanks,
Andrew