Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Jan Kratochvil <jan.kratochvil@redhat.com>
To: Pedro Alves <palves@redhat.com>
Cc: gdb-patches@sourceware.org, Victor Leschuk <vleschuk@accesssoftek.com>
Subject: Re: [pushed] Re: [PATCH 1/6] Code cleanup: C++ify .gdb_index producer.
Date: Sun, 18 Jun 2017 16:50:00 -0000	[thread overview]
Message-ID: <20170618165045.GA6469@host1.jankratochvil.net> (raw)
In-Reply-To: <8efc0742-1014-4fe0-6948-f40a9c5c4975@redhat.com>

On Mon, 12 Jun 2017 18:08:07 +0200, Pedro Alves wrote:
> Note that this patch as is actually causes a ~10% performance
> drop in .gdb_index generation that can be significant when
> running gdb index on big binaries or in a a batch of binaries.

gdb-add-index is being used in Koji builds where it takes minutes to build
even a trivial package.

Otherwise I do not see why anyone would use it for C++ programs
edit-compile-debug cycles where the speed matters:

0m 5.389s -fuse-ld=gold -Wl,--gdb-index,--build-id
0m 3.564s -fuse-ld=gold             -Wl,--build-id
   1.825s = --gdb-index part of gold
0m19.401s (GNU ld)                  -Wl,--build-id
0m 4.364s gdb-add-index
  23.765s = GNU ld + gdb-add-index
23.765 / 5.389 = 4.4099090740397102

GNU ld is too slow and then gold can already produce .gdb_index along its
linking in just negligible time.

Just gold is not a default linker as discussed in:
	[PATCH] [BZ #14995] glibc fails to build if gold is the default linker, even if ld.bfd is available
	https://sourceware.org/ml/libc-alpha/2015-03/msg00410.html
and in my mail internally in Red Hat:
	Subject: [debug-list] DWARF5 .debug_names vs. .gdb_index vs. two lds
	Message-ID: <20161027162728.GA6631@host1.jankratochvil.net>

Although nobody invests time to fix gold to finally switch to it.

But then the currently missing support for DWO in .debug_names seems to be
pointless to implement into GDB as DWO is used only for better
edit-compile-debug cycle speed but then one cannot use GNU ld and so DWO
.debug_names support should be implemented rather into gold.


Jan


  parent reply	other threads:[~2017-06-18 16:50 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-26 18:25 [PATCH 0/6] DWARF-5: .debug_names index Jan Kratochvil
2017-05-26 18:25 ` [PATCH 1/6] Code cleanup: C++ify .gdb_index producer Jan Kratochvil
2017-06-12 16:08   ` [pushed] " Pedro Alves
2017-06-12 16:14     ` [PATCH 3/6] Code cleanup: dwarf2read.c: Add data_buf::append_uint Pedro Alves
2017-06-12 16:14     ` [PATCH 1/6] Code cleanup: dwarf2read.c:uniquify_cu_indices: Use std::unique Pedro Alves
2017-06-12 16:14     ` [PATCH 4/6] .gdb_index prod perf regression: find before insert in unordered_map Pedro Alves
2017-06-12 16:14     ` [PATCH 6/6] .gdb_index prod perf regression: mapped_symtab now vector of values Pedro Alves
2017-06-12 16:14     ` [PATCH 2/6] Code cleanup: dwarf2read.c: Eliminate ::file_write Pedro Alves
2017-06-17 17:35       ` Jan Kratochvil
2017-06-19  9:26         ` Pedro Alves
2017-06-18 18:36       ` Regression: " Jan Kratochvil
2017-06-19  9:27         ` Pedro Alves
2017-06-19  9:39           ` Jan Kratochvil
2017-06-19  9:47             ` Pedro Alves
2017-06-19 10:03               ` Jan Kratochvil
2017-06-19 10:35                 ` Pedro Alves
2017-06-19 12:06                 ` Yao Qi
2017-06-19 12:26                   ` Jan Kratochvil
2017-06-12 16:18     ` [pushed] Re: [PATCH 1/6] Code cleanup: C++ify .gdb_index producer Pedro Alves
2017-06-12 16:19     ` [PATCH 5/6] .gdb_index prod perf regression: Estimate size of psyms_seen Pedro Alves
2017-06-18 14:25     ` [pushed] Re: [PATCH 1/6] Code cleanup: C++ify .gdb_index producer Jan Kratochvil
2017-06-18 15:12       ` Eli Zaretskii
2017-06-19 11:50         ` [pushed] .gdb_index writer: close the file before unlinking it (Re: [pushed] Re: [PATCH 1/6] Code cleanup: C++ify .gdb_index producer.) Pedro Alves
2017-06-18 16:50     ` Jan Kratochvil [this message]
2017-05-26 18:25 ` [PATCH 2/6] cc-with-tweaks.sh: Use gdb-add-index.sh Jan Kratochvil
2017-05-26 18:26 ` [PATCH 4/6] Code cleanup: dwarf2_initialize_objfile return value Jan Kratochvil
2017-05-26 18:26 ` [PATCH 3/6] DWARF-5: .debug_names index producer Jan Kratochvil
2017-06-09  5:58   ` [PATCH 3.1/6] " Jan Kratochvil
2017-05-26 18:26 ` [PATCH 5/6] Refactor: Move some generic code out of .gdb_index code Jan Kratochvil
2017-05-26 18:26 ` [PATCH 6/6] DWARF-5: .debug_names index consumer Jan Kratochvil
2017-06-18 19:37 ` obsolete: [PATCH 0/6] DWARF-5: .debug_names index Jan Kratochvil

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20170618165045.GA6469@host1.jankratochvil.net \
    --to=jan.kratochvil@redhat.com \
    --cc=gdb-patches@sourceware.org \
    --cc=palves@redhat.com \
    --cc=vleschuk@accesssoftek.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox