From: "Ulrich Weigand" <uweigand@de.ibm.com>
To: tom@tromey.com (Tom Tromey)
Cc: palves@redhat.com (Pedro Alves),
gdb-patches@sourceware.org, tom@tromey.com,
simon.marchi@ericsson.com
Subject: [pushed] Re: [RFA 1/6] Use std::vector in end_symtab_get_static_block
Date: Tue, 24 Oct 2017 14:41:00 -0000 [thread overview]
Message-ID: <20171024144059.41BA2D80756@oc3748833570.ibm.com> (raw)
In-Reply-To: <87lgk03b1f.fsf@tromey.com> from "Tom Tromey" at Oct 24, 2017 07:55:40 AM
Tom Tromey wrote:
> >>>>> "Ulrich" == Ulrich Weigand <uweigand@de.ibm.com> writes:
>
> Ulrich> Now, the reason why we re-sort pending_blocks here is to allow
> Ulrich> for object code reordering (hot/cold sections and the like).
> Ulrich> But this type of reordering never actually affects inline function
> Ulrich> relationships. So it may be that just using stable_sort here is
> Ulrich> actually the *correct* fix ...
>
> I tend to think so as well, for the reasons you mentioned.
> glibc doesn't claim that qsort is stable, but maybe it is in practice
> sometimes?
I just had a look at the glibc qsort implementation. It actually uses
(in the common case) a *merge sort* into a temporary buffer, which is
always stable. The GNU libstdc++ std::sort on the other hand seems to
use an in-place quicksort algorithm, which is generally not stable.
This would explain the difference.
I've now pushed the patch in the form below.
Thanks,
Ulrich
gdb/ChangeLog:
2017-10-24 Ulrich Weigand <uweigand@de.ibm.com>
* buildsym.c (end_symtab_get_static_block): Use std::stable_sort.
diff --git a/gdb/buildsym.c b/gdb/buildsym.c
index c556ac1..07bfbd5 100644
--- a/gdb/buildsym.c
+++ b/gdb/buildsym.c
@@ -1249,12 +1249,14 @@ end_symtab_get_static_block (CORE_ADDR end_addr, int expandable, int required)
for (pb = pending_blocks; pb != NULL; pb = pb->next)
barray.push_back (pb->block);
- std::sort (barray.begin (), barray.end (),
- [] (const block *a, const block *b)
- {
- /* Sort blocks in descending order. */
- return BLOCK_START (a) > BLOCK_START (b);
- });
+ /* Sort blocks by start address in descending order. Blocks with the
+ same start address must remain in the original order to preserve
+ inline function caller/callee relationships. */
+ std::stable_sort (barray.begin (), barray.end (),
+ [] (const block *a, const block *b)
+ {
+ return BLOCK_START (a) > BLOCK_START (b);
+ });
int i = 0;
for (pb = pending_blocks; pb != NULL; pb = pb->next)
--
Dr. Ulrich Weigand
GNU/Linux compilers and toolchain
Ulrich.Weigand@de.ibm.com
next prev parent reply other threads:[~2017-10-24 14:41 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-10-16 3:04 [RFA 0/6] more cleanup removals Tom Tromey
2017-10-16 3:04 ` [RFA 2/6] Remove some cleanups from probe.c Tom Tromey
2017-10-16 20:26 ` Simon Marchi
2017-10-16 3:04 ` [RFA 5/6] Return unique_xmalloc_ptr from target_read_stralloc Tom Tromey
2017-10-16 21:02 ` Simon Marchi
2017-10-16 3:04 ` [RFA 4/6] Simple cleanup removals in remote.c Tom Tromey
2017-10-16 20:43 ` Simon Marchi
2017-10-16 21:14 ` Tom Tromey
2017-10-16 21:37 ` Simon Marchi
2017-10-16 21:50 ` Tom Tromey
2017-10-16 22:35 ` Pedro Alves
2017-10-16 23:00 ` Tom Tromey
2017-10-16 23:34 ` [PATCH 1/2] Introduce string_appendf/string_vappendf (Re: [RFA 4/6] Simple cleanup removals in remote.c) Pedro Alves
2017-10-19 3:11 ` Simon Marchi
2017-10-19 3:13 ` Simon Marchi
2017-10-30 0:16 ` Simon Marchi
2017-10-30 11:48 ` Pedro Alves
2017-10-16 23:38 ` [PATCH 2/2] remote.c, QCatchSyscalls: Build std::string instead of unique_xmalloc_ptr " Pedro Alves
2017-10-19 3:17 ` Simon Marchi
2017-10-30 11:49 ` Pedro Alves
2017-10-16 3:04 ` [RFA 3/6] Remove cleanup from ppc-linux-nat.c Tom Tromey
2017-10-16 20:28 ` Simon Marchi
2017-10-16 3:04 ` [RFA 1/6] Use std::vector in end_symtab_get_static_block Tom Tromey
2017-10-16 20:18 ` Simon Marchi
2017-10-16 21:59 ` Tom Tromey
2017-10-20 15:33 ` Ulrich Weigand
2017-10-20 16:47 ` Pedro Alves
2017-10-23 16:16 ` Ulrich Weigand
2017-10-24 13:55 ` Tom Tromey
2017-10-24 14:41 ` Ulrich Weigand [this message]
2017-10-16 3:04 ` [RFA 6/6] Return unique_xmalloc_ptr from target_fileio_read_stralloc Tom Tromey
2017-10-16 21:07 ` Simon Marchi
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=20171024144059.41BA2D80756@oc3748833570.ibm.com \
--to=uweigand@de.ibm.com \
--cc=gdb-patches@sourceware.org \
--cc=palves@redhat.com \
--cc=simon.marchi@ericsson.com \
--cc=tom@tromey.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).