From: Nathan Chancellor <nathan@kernel.org>
To: Nick Desaulniers <ndesaulniers@google.com>
Cc: Masahiro Yamada <masahiroy@kernel.org>,
Fangrui Song <maskray@google.com>,
Sedat Dilek <sedat.dilek@dhl.com>,
Michal Marek <michal.lkml@markovi.net>,
Nick Terrell <terrelln@fb.com>, Tom Rix <trix@redhat.com>,
Andrew Morton <akpm@linux-foundation.org>,
"Peter Zijlstra (Intel)" <peterz@infradead.org>,
David Gow <davidgow@google.com>,
Kees Cook <keescook@chromium.org>,
Josh Poimboeuf <jpoimboe@kernel.org>,
Dan Williams <dan.j.williams@intel.com>,
Miguel Ojeda <ojeda@kernel.org>,
Isabella Basso <isabbasso@riseup.net>,
Vlastimil Babka <vbabka@suse.cz>,
Rasmus Villemoes <linux@rasmusvillemoes.dk>,
linux-kernel@vger.kernel.org, linux-kbuild@vger.kernel.org,
llvm@lists.linux.dev
Subject: Re: [PATCH] Makefile.debug: support for -gz=zstd
Date: Thu, 20 Oct 2022 11:39:50 -0700 [thread overview]
Message-ID: <Y1GV9sHyODVmBbFW@dev-arch.thelio-3990X> (raw)
In-Reply-To: <20221020175655.1660864-1-ndesaulniers@google.com>
On Thu, Oct 20, 2022 at 10:56:49AM -0700, Nick Desaulniers wrote:
> Make DEBUG_INFO_COMPRESSED a choice; DEBUG_INFO_UNCOMPRESSED is the
> default, DEBUG_INFO_COMPRESSED uses zlib, DEBUG_INFO_COMPRESSED_ZSTD
> uses zstd.
>
> Some quick N=1 measurements with du, /usr/bin/time -v, and bloaty:
>
> clang-16, x86_64 defconfig plus
> CONFIG_DEBUG_INFO=y CONFIG_DEBUG_INFO_UNCOMPRESSED=y:
> Elapsed (wall clock) time (h:mm:ss or m:ss): 0:55.43
> 488M vmlinux
> 27.6% 136Mi 0.0% 0 .debug_info
> 6.1% 30.2Mi 0.0% 0 .debug_str_offsets
> 3.5% 17.2Mi 0.0% 0 .debug_line
> 3.3% 16.3Mi 0.0% 0 .debug_loclists
> 0.9% 4.62Mi 0.0% 0 .debug_str
>
> clang-16, x86_64 defconfig plus
> CONFIG_DEBUG_INFO=y CONFIG_DEBUG_INFO_COMPRESSED=y (zlib):
> Elapsed (wall clock) time (h:mm:ss or m:ss): 1:00.35
> 385M vmlinux
> 21.8% 85.4Mi 0.0% 0 .debug_info
> 2.1% 8.26Mi 0.0% 0 .debug_str_offsets
> 2.1% 8.24Mi 0.0% 0 .debug_loclists
> 1.9% 7.48Mi 0.0% 0 .debug_line
> 0.5% 1.94Mi 0.0% 0 .debug_str
>
> clang-16, x86_64 defconfig plus
> CONFIG_DEBUG_INFO=y CONFIG_DEBUG_INFO_COMPRESSED_ZSTD=y (zstd):
> Elapsed (wall clock) time (h:mm:ss or m:ss): 0:59.69
> 373M vmlinux
> 21.4% 81.4Mi 0.0% 0 .debug_info
> 2.3% 8.85Mi 0.0% 0 .debug_loclists
> 1.5% 5.71Mi 0.0% 0 .debug_line
> 0.5% 1.95Mi 0.0% 0 .debug_str_offsets
> 0.4% 1.62Mi 0.0% 0 .debug_str
>
> That's only a 3.11% overall binary size savings over zlib, but at no
> performance regression.
>
> Link: https://maskray.me/blog/2022-09-09-zstd-compressed-debug-sections
> Link: https://maskray.me/blog/2022-01-23-compressed-debug-sections
> Suggested-by: Sedat Dilek (DHL Supply Chain) <sedat.dilek@dhl.com>
> Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
One small comment below.
> ---
> lib/Kconfig.debug | 26 +++++++++++++++++++++++++-
> scripts/Makefile.debug | 4 ++++
> 2 files changed, 29 insertions(+), 1 deletion(-)
>
> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> index 3fc7abffc7aa..4085ac77dc12 100644
> --- a/lib/Kconfig.debug
> +++ b/lib/Kconfig.debug
> @@ -312,8 +312,22 @@ config DEBUG_INFO_REDUCED
> DEBUG_INFO build and compile times are reduced too.
> Only works with newer gcc versions.
>
> +choice
> + prompt "Compressed Debug information"
> + depends on DEBUG_KERNEL
I think you can drop this depends. The entire block is in an
'if DEBUG_INFO', which can only be true if CONFIG_DEBUG_KERNEL is set
because of the dependencies of the "Debug information" prompt above this
file, which is how CONFIG_DEBUG_INFO is selected.
> + help
> + Compress the resulting debug info. Results in smaller debug info sections,
> + but requires that consumers are able to decompress the results.
> +
> + If unsure, choose DEBUG_INFO_UNCOMPRESSED.
> +
> +config DEBUG_INFO_UNCOMPRESSED
> + bool "Don't compress debug information"
> + help
> + Don't compress debug info sections.
> +
> config DEBUG_INFO_COMPRESSED
> - bool "Compressed debugging information"
> + bool "Compress debugging information with zlib"
> depends on $(cc-option,-gz=zlib)
> depends on $(ld-option,--compress-debug-sections=zlib)
> help
> @@ -327,6 +341,16 @@ config DEBUG_INFO_COMPRESSED
> preferable to setting $KDEB_COMPRESS to "none" which would be even
> larger.
>
> +config DEBUG_INFO_COMPRESSED_ZSTD
> + bool "Compress debugging information with zstd"
> + depends on $(cc-option,-gz=zstd)
> + depends on $(ld-option,--compress-debug-sections=zstd)
> + help
> + Compress the debug information using zstd. Requires GCC 13.0+ or Clang
> + 16.0+, binutils 2.40+, and zstd.
> +
> +endchoice # "Compressed Debug information"
> +
> config DEBUG_INFO_SPLIT
> bool "Produce split debuginfo in .dwo files"
> depends on $(cc-option,-gsplit-dwarf)
> diff --git a/scripts/Makefile.debug b/scripts/Makefile.debug
> index 332c486f705f..8ac3379d2255 100644
> --- a/scripts/Makefile.debug
> +++ b/scripts/Makefile.debug
> @@ -31,6 +31,10 @@ ifdef CONFIG_DEBUG_INFO_COMPRESSED
> DEBUG_CFLAGS += -gz=zlib
> KBUILD_AFLAGS += -gz=zlib
> KBUILD_LDFLAGS += --compress-debug-sections=zlib
> +else ifdef CONFIG_DEBUG_INFO_COMPRESSED_ZSTD
> +DEBUG_CFLAGS += -gz=zstd
> +KBUILD_AFLAGS += -gz=zstd
> +KBUILD_LDFLAGS += --compress-debug-sections=zstd
> endif
>
> KBUILD_CFLAGS += $(DEBUG_CFLAGS)
> --
> 2.38.0.135.g90850a2211-goog
>
next prev parent reply other threads:[~2022-10-20 18:39 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-20 17:56 [PATCH] Makefile.debug: support for -gz=zstd Nick Desaulniers
2022-10-20 18:39 ` Nathan Chancellor [this message]
2022-10-23 15:45 ` Sedat Dilek
2022-10-24 17:44 ` Nick Desaulniers
2022-10-25 17:08 ` AW: " Sedat Dilek (DHL Supply Chain)
2022-10-31 18:49 ` Nick Desaulniers
2022-11-01 5:57 ` Masahiro Yamada
2022-11-07 18:01 ` [PATCH v2] " Nick Desaulniers
2022-11-08 4:43 ` Nicolas Schier
2022-11-08 23:41 ` Sedat Dilek
2022-11-10 19:59 ` [PATCH v3] " Nick Desaulniers
2022-11-13 10:55 ` Masahiro Yamada
2023-01-03 19:39 ` Sedat Dilek
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=Y1GV9sHyODVmBbFW@dev-arch.thelio-3990X \
--to=nathan@kernel.org \
--cc=akpm@linux-foundation.org \
--cc=dan.j.williams@intel.com \
--cc=davidgow@google.com \
--cc=isabbasso@riseup.net \
--cc=jpoimboe@kernel.org \
--cc=keescook@chromium.org \
--cc=linux-kbuild@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux@rasmusvillemoes.dk \
--cc=llvm@lists.linux.dev \
--cc=masahiroy@kernel.org \
--cc=maskray@google.com \
--cc=michal.lkml@markovi.net \
--cc=ndesaulniers@google.com \
--cc=ojeda@kernel.org \
--cc=peterz@infradead.org \
--cc=sedat.dilek@dhl.com \
--cc=terrelln@fb.com \
--cc=trix@redhat.com \
--cc=vbabka@suse.cz \
/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