From: David Blaikie via Gdb <gdb@sourceware.org>
To: "R. Diez" <rdiezmail-binutils@yahoo.de>
Cc: "gdb@sourceware.org" <gdb@sourceware.org>
Subject: Re: Greatly increased GDB memory and CPU usage with newest embedded ARM toolchain
Date: Sun, 18 Apr 2021 21:57:32 -0700 [thread overview]
Message-ID: <CAENS6EsorosiyktZrkJLXEsUFGsHSVSPfWzxxTvDw=wc_Fdo7A@mail.gmail.com> (raw)
In-Reply-To: <95f8714b-a105-8759-dcc4-73122ee8bcf2@yahoo.de>
4My high water mark reported my massif for the non-lto build was 197MB -
most of that seems to come from decoding the macro information
(dwarf_decode_macros). You could try dropping down from -g3 to -g2 to see
if that helps. (I realize the linux kernel uses a bunch of macros and
benefits from debug info for macros, but it'd at least help isolate the
problem - might help clarify whether the lto case is related or. not)
On Sun, Apr 18, 2021 at 10:02 AM R. Diez <rdiezmail-binutils@yahoo.de>
wrote:
>
> > [...]
> > & yeah, that does seem like quite a bit of RAM usage for a relatively
> > small amount of debug info. Though I'm not a regular/frequent gdb
> > developer, so I don't have any particular insight there - if no one
> > else chimes in, might be worth running valgrind --tool=massif to get a
> > memory profile, might point to what part of gdb is using all the RAM.
>
> I am worried that I may not be building the toolchain and/or GDB correctly.
>
> Could you do the following test for me, just to confirm that you are
> indeed seeing such a high memory consumption?
>
> Could you build GDB 10.1 for ARM? This is how I am configuring it for
> cross-debugging (somewhat simplified):
>
> configure \
> CFLAGS="-g0 -O3 -flto=9 -march=native" \
> CXXFLAGS="-g0 -O3 -flto=9 -march=native" \
> --target=arm-none-eabi
>
> Maybe you want to build it without -O3 and LTO, just using the defaults. I
> have been using those flags for years with previous versions without any
> issues.
>
> Then load one of the .elf files in the attachment like this. There is no
> need to have any ARM CPU available:
>
> ./arm-none-eabi-gdb firmware-debug-non-lto.elf
>
> At this point, GDB should be using less than 15 MiB of RAM.
>
> Now issue this GDB command:
>
> print StartOfUserCode
>
> You should see an output like this:
>
> $1 = {void (void)} 0x866d8 <StartOfUserCode()>
>
> Did that take more than 1 second? How much RAM is your GDB using now?
>
> If you repeat that with firmware-release-lto.elf , how much memory is your
> GDB using? Be prepared to kill it before it breaks your system though. I
> normally kill it after a few seconds when it reaches 2 GiB of RAM.
>
> Thanks in advance,
> rdiez
>
next prev parent reply other threads:[~2021-04-19 4:57 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <537840208.5089288.1618695042450.ref@mail.yahoo.com>
2021-04-17 21:30 ` R. Diez via Gdb
2021-04-17 22:15 ` David Blaikie via Gdb
2021-04-18 8:16 ` R. Diez via Gdb
2021-04-18 15:41 ` David Blaikie via Gdb
[not found] ` <95f8714b-a105-8759-dcc4-73122ee8bcf2@yahoo.de>
2021-04-19 4:57 ` David Blaikie via Gdb [this message]
2021-04-19 5:10 ` David Blaikie via Gdb
2021-04-19 16:08 ` R. Diez via Gdb
2021-04-19 7:06 ` R. Diez via Gdb
2021-04-19 1:07 ` Simon Marchi via Gdb
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='CAENS6EsorosiyktZrkJLXEsUFGsHSVSPfWzxxTvDw=wc_Fdo7A@mail.gmail.com' \
--to=gdb@sourceware.org \
--cc=dblaikie@gmail.com \
--cc=rdiezmail-binutils@yahoo.de \
/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