public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Andi Kleen <ak@linux.intel.com>
To: Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>
Cc: "Jiri Slaby (SUSE)" <jirislaby@kernel.org>,
	linux-kernel@vger.kernel.org, Andy Lutomirski <luto@kernel.org>,
	Martin Liska <mliska@suse.cz>, Jiri Slaby <jslaby@suse.cz>
Subject: Re: [PATCH 18/46] entry, lto: Mark raw_irqentry_exit_cond_resched() as __visible
Date: Thu, 17 Nov 2022 14:07:23 -0800	[thread overview]
Message-ID: <289e03d2-be50-4249-343a-75dae302b0e5@linux.intel.com> (raw)
In-Reply-To: <Y3XzkxNVTvdB4a/1@hirez.programming.kicks-ass.net>


> I still don't understand any of it -- this symbol is not static (and
> thus lives in the global namespace and it's name must not be mangled
> lest it breaks ABI), this symbol has it's address taken, so it must not
> be eliminated.

It's not eliminated, but is still manged because gcc turns it into 
static due to

-fwhole-program. Maybe this could avoided in gcc, but at least that's 
what it does currently.

I believe disabling -fwhole-program would likely avoid it, but it would 
also prevent some code

transformations because gcc would need to assume that every function can 
be called by

someone it doesn't see.

> WTF does this crazy LTO thing require __visible on it?
>
> The original Changelog babbles something about multiple object files,
> which doesn't make sense either, there is only a single object file with
> LTO -- that's sort of the whole point. The translation unit output
> becomes some intermediate gunk -- to be used as input for the LTO pass,
> but it is not an ELF object file.
>
> The linker takes all these intermediate files, does the global
> optimization thing and then generates a real ELF object file.

That would be a single threaded very very slow global compilation. 
Instead gcc WHOPR uses

partitioning to generate smaller units that can be compiled in parallel 
based on their call dependencies,

and these use different object files from the individual assembler 
invocations.

>
> Anyway; I think we can drop all this crazy on the floor again, since per
> the 0/n (which I didn't get) there isn't any actual benefit from using
> GCC-LTO, so why should we bother with all this ugly.

At least in the past it generated smaller kernels for small configurations.

One benefit that wasn't mentioned is doing type and other checks (e.g. 
constant propagation

through inlining) across files.

In general LTO gives the compiler a lot more freedom to optimize code, 
so even if it's not quite there

yet I think it's beneficial to let users play around with it and see if 
they can get benefits.



-Andi


  reply	other threads:[~2022-11-17 22:08 UTC|newest]

Thread overview: 87+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-14 11:42 [PATCH 00/46] gcc-LTO support for the kernel Jiri Slaby (SUSE)
2022-11-14 11:42 ` [PATCH 01/46] x86/boot: robustify calling startup_{32,64}() from the decompressor code Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 02/46] kbuild: pass jobserver to cmd_ld_vmlinux.o Jiri Slaby (SUSE)
2022-11-14 17:57   ` Masahiro Yamada
2022-11-15  6:36     ` Jiri Slaby
2022-11-14 11:43 ` [PATCH 03/46] kbuild: lto: preserve MAKEFLAGS for module linking Jiri Slaby (SUSE)
2022-11-14 18:02   ` Masahiro Yamada
2022-11-14 11:43 ` [PATCH 04/46] compiler.h: introduce __visible_on_lto Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 05/46] compiler.h: introduce __global_on_lto Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 06/46] Compiler Attributes, lto: introduce __noreorder Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 07/46] tracepoint, lto: Mark static call functions as __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 08/46] static_call, lto: Mark static keys " Jiri Slaby (SUSE)
2022-11-14 15:51   ` Peter Zijlstra
2022-11-14 18:52     ` Josh Poimboeuf
2022-11-14 20:34     ` Andi Kleen
2022-11-17  8:24       ` Peter Zijlstra
2022-11-14 18:57   ` Josh Poimboeuf
2022-11-14 11:43 ` [PATCH 09/46] static_call, lto: Mark static_call_return0() " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 10/46] static_call, lto: Mark func_a() as __visible_on_lto Jiri Slaby (SUSE)
2022-11-14 15:54   ` Peter Zijlstra
2022-11-14 20:29     ` Andi Kleen
2022-11-14 11:43 ` [PATCH 11/46] x86/alternative, lto: Mark int3_*() as global and __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 12/46] x86/paravirt, lto: Mark native_steal_clock() as __visible_on_lto Jiri Slaby (SUSE)
2022-11-14 15:58   ` Peter Zijlstra
2022-11-14 11:43 ` [PATCH 13/46] x86/preempt, lto: Mark preempt_schedule_*thunk() as __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 14/46] x86/sev, lto: Mark cpuid_table_copy as __visible_on_lto Jiri Slaby (SUSE)
2022-11-14 16:02   ` Peter Zijlstra
2022-11-14 11:43 ` [PATCH 15/46] x86/xen, lto: Mark xen_vcpu_stolen() as __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 16/46] x86, lto: Mark gdt_page and native_sched_clock() " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 17/46] amd, lto: Mark amd pmu and pstate functions as __visible_on_lto Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 18/46] entry, lto: Mark raw_irqentry_exit_cond_resched() as __visible Jiri Slaby (SUSE)
2022-11-16 23:30   ` Thomas Gleixner
2022-11-17  8:40     ` Peter Zijlstra
2022-11-17 22:07       ` Andi Kleen [this message]
2022-11-18  1:28         ` Thomas Gleixner
2022-11-19  0:50           ` Andi Kleen
2022-11-19  8:50             ` Thomas Gleixner
2022-11-22  9:32     ` Jiri Slaby
2022-11-14 11:43 ` [PATCH 19/46] export, lto: Mark __kstrtab* in EXPORT_SYMBOL() as global and __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 20/46] softirq, lto: Mark irq_enter/exit_rcu() as __visible Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 21/46] btf, lto: pass scope as strings Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 22/46] btf, lto: Make all BTF IDs global on LTO Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 23/46] init.h, lto: mark initcalls as __noreorder Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 24/46] bpf, lto: mark interpreter jump table " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 25/46] sched, lto: mark sched classes " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 26/46] x86/apic, lto: Mark apic_driver*() " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 27/46] linkage, lto: use C version for SYSCALL_ALIAS() / cond_syscall() Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 28/46] scripts, lto: re-add gcc-ld Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 29/46] scripts, lto: use CONFIG_LTO for many LTO specific actions Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 30/46] Kbuild, lto: Add Link Time Optimization support Jiri Slaby (SUSE)
2022-11-14 18:55   ` Josh Poimboeuf
2022-11-15 13:31     ` Martin Liška
2022-11-14 11:43 ` [PATCH 31/46] x86/purgatory, lto: Disable gcc LTO for purgatory Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 32/46] x86/realmode, lto: Disable gcc LTO for real mode code Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 33/46] x86/vdso, lto: Disable gcc LTO for the vdso Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 34/46] scripts, lto: disable gcc LTO for some mod sources Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 35/46] Kbuild, lto: disable gcc LTO for bounds+asm-offsets Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 36/46] lib/string, lto: disable gcc LTO for string.o Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 37/46] Compiler attributes, lto: disable __flatten with LTO Jiri Slaby (SUSE)
2022-11-14 17:01   ` Miguel Ojeda
2022-11-14 11:43 ` [PATCH 38/46] Kbuild, lto: don't include weak source file symbols in System.map Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 39/46] x86, lto: Disable relative init pointers with gcc LTO Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 40/46] x86/livepatch, lto: Disable live patching " Jiri Slaby (SUSE)
2022-11-14 19:07   ` Josh Poimboeuf
2022-11-14 20:28     ` Andi Kleen
2022-11-14 22:00       ` Josh Poimboeuf
2022-11-15 13:32         ` Martin Liška
2022-11-17 20:00   ` Song Liu
2022-11-14 11:43 ` [PATCH 41/46] x86/lib, lto: Mark 32bit mem{cpy,move,set} as __used Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 42/46] mm/kasan, lto: Mark kasan " Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 43/46] scripts, lto: check C symbols for modversions Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 44/46] scripts/bloat-o-meter, lto: handle gcc LTO Jiri Slaby (SUSE)
2022-11-14 11:43 ` [PATCH 45/46] kasan, lto: remove extra BUILD_BUG() in memory_is_poisoned Jiri Slaby (SUSE)
2022-11-26 17:07   ` Andrey Konovalov
2022-11-14 11:43 ` [PATCH 46/46] x86, lto: Finally enable gcc LTO for x86 Jiri Slaby (SUSE)
2022-11-14 11:56 ` [PATCH 00/46] gcc-LTO support for the kernel Ard Biesheuvel
2022-11-14 12:04   ` Jiri Slaby
2022-11-14 19:40 ` Ard Biesheuvel
2022-11-17  8:28   ` Peter Zijlstra
2022-11-17  8:50     ` Richard Biener
2022-11-17 11:42       ` Peter Zijlstra
2022-11-17 11:49         ` Ard Biesheuvel
2022-11-17 13:55           ` Richard Biener
2022-11-17 14:32             ` Peter Zijlstra
2022-11-17 14:40               ` Richard Biener
2022-11-17 15:15             ` Ard Biesheuvel
2022-11-17 11:48       ` Thomas Gleixner

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=289e03d2-be50-4249-343a-75dae302b0e5@linux.intel.com \
    --to=ak@linux.intel.com \
    --cc=jirislaby@kernel.org \
    --cc=jslaby@suse.cz \
    --cc=linux-kernel@vger.kernel.org \
    --cc=luto@kernel.org \
    --cc=mliska@suse.cz \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.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