All of lore.kernel.org
 help / color / mirror / Atom feed
From: Nathan Chancellor <nathan@kernel.org>
To: Nick Desaulniers <ndesaulniers@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Arnd Bergmann <arnd@arndb.de>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Tom Rix <trix@redhat.com>, Xiaoming Ni <nixiaoming@huawei.com>,
	Luis Chamberlain <mcgrof@kernel.org>,
	Andi Kleen <ak@linux.intel.com>,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	llvm@lists.linux.dev, Christoph Hellwig <hch@lst.de>,
	Eric Dumazet <edumazet@google.com>,
	"Darrick J. Wong" <darrick.wong@oracle.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [PATCH v2] fs/select: mark do_select noinline_for_stack
Date: Fri, 14 Oct 2022 15:39:35 -0700	[thread overview]
Message-ID: <Y0nlJ6whtJuZddjr@dev-arch.thelio-3990X> (raw)
In-Reply-To: <20221011205547.14553-1-ndesaulniers@google.com>

On Tue, Oct 11, 2022 at 01:55:47PM -0700, Nick Desaulniers wrote:
> Effectively a revert of
> commit ad312f95d41c ("fs/select: avoid clang stack usage warning")
> 
> Various configs can still push the stack useage of core_sys_select()
> over the CONFIG_FRAME_WARN threshold (1024B on 32b targets).
> 
>   fs/select.c:619:5: error: stack frame size of 1048 bytes in function
>   'core_sys_select' [-Werror,-Wframe-larger-than=]
> 
> core_sys_select() has a large stack allocation for `stack_fds` where it
> tries to do something equivalent to "small string optimization" to
> potentially avoid a kmalloc.
> 
> core_sys_select() calls do_select() which has another potentially large
> stack allocation, `table`. Both of these values depend on
> FRONTEND_STACK_ALLOC.
> 
> Mix those two large allocation with register spills which are
> exacerbated by various configs and compiler versions and we can just
> barely exceed the 1024B limit.
> 
> Rather than keep trying to find the right value of MAX_STACK_ALLOC or
> FRONTEND_STACK_ALLOC, mark do_select() as noinline_for_stack.
> 
> The intent of FRONTEND_STACK_ALLOC is to help potentially avoid a
> dynamic memory allocation. In that spirit, restore the previous
> threshold but separate the stack frames.
> 
> Many tests of various configs for different architectures and various
> versions of GCC were performed; do_select() was never inlined into
> core_sys_select() or compat_core_sys_select(). The kernel is built with
> the GCC specific flag `-fconserve-stack` which can limit inlining
> depending on per-target thresholds of callee stack size, which helps
> avoid the issue when using GCC. Clang is being more aggressive and not
> considering the stack size when decided whether to inline or not. We may
> consider using the clang-16+ flag `-finline-max-stacksize=` in the
> future.
> 
> Link: https://lore.kernel.org/lkml/20221006222124.aabaemy7ofop7ccz@google.com/
> Fixes: ad312f95d41c ("fs/select: avoid clang stack usage warning")
> Suggested-by: Arnd Bergmann <arnd@arndb.de>
> Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>

Tested-by: Nathan Chancellor <nathan@kernel.org>

> ---
> Changes v1 -> v2:
> * Drop the 32b specific guard, since I could reproduce the no-inlining
>   w/ aarch64-linux-gnu-gcc-10 ARCH=arm64 defconfig, and per Arnd.
> * Drop references to 32b in commit message.
> * Add new paragraph in commit message at the end about -fconserve-stack
>   and -finline-max-stacksize=.
> * s/malloc/kmalloc/ in commit message.
> 
>  fs/select.c          | 1 +
>  include/linux/poll.h | 4 ----
>  2 files changed, 1 insertion(+), 4 deletions(-)
> 
> diff --git a/fs/select.c b/fs/select.c
> index 0ee55af1a55c..794e2a91b1fa 100644
> --- a/fs/select.c
> +++ b/fs/select.c
> @@ -476,6 +476,7 @@ static inline void wait_key_set(poll_table *wait, unsigned long in,
>  		wait->_key |= POLLOUT_SET;
>  }
>  
> +noinline_for_stack
>  static int do_select(int n, fd_set_bits *fds, struct timespec64 *end_time)
>  {
>  	ktime_t expire, *to = NULL;
> diff --git a/include/linux/poll.h b/include/linux/poll.h
> index a9e0e1c2d1f2..d1ea4f3714a8 100644
> --- a/include/linux/poll.h
> +++ b/include/linux/poll.h
> @@ -14,11 +14,7 @@
>  
>  /* ~832 bytes of stack space used max in sys_select/sys_poll before allocating
>     additional memory. */
> -#ifdef __clang__
> -#define MAX_STACK_ALLOC 768
> -#else
>  #define MAX_STACK_ALLOC 832
> -#endif
>  #define FRONTEND_STACK_ALLOC	256
>  #define SELECT_STACK_ALLOC	FRONTEND_STACK_ALLOC
>  #define POLL_STACK_ALLOC	FRONTEND_STACK_ALLOC
> -- 
> 2.38.0.rc2.412.g84df46c1b4-goog
> 
> 

  reply	other threads:[~2022-10-14 22:39 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-03-07  9:01 [PATCH] fs/select: avoid clang stack usage warning Arnd Bergmann
2019-03-07 16:19 ` Andi Kleen
2022-10-06 22:21 ` Nick Desaulniers
2022-10-07  8:28   ` Arnd Bergmann
2022-10-07 19:04     ` Nick Desaulniers
2022-10-07 20:11       ` [PATCH] fs/select: mark do_select noinline_for_stack for 32b Nick Desaulniers
2022-10-10  7:44         ` Christoph Hellwig
2022-10-10 16:42           ` Nick Desaulniers
2022-10-10 19:55         ` Arnd Bergmann
2022-10-13 18:52         ` kernel test robot
2022-10-07 21:42       ` [PATCH] fs/select: avoid clang stack usage warning Arnd Bergmann
2022-10-07 22:54         ` Nick Desaulniers
2022-10-10 18:40           ` Nick Desaulniers
2022-10-11 12:46             ` Arnd Bergmann
2022-10-11 20:55               ` [PATCH v2] fs/select: mark do_select noinline_for_stack Nick Desaulniers
2022-10-14 22:39                 ` Nathan Chancellor [this message]
2022-10-07 23:17         ` [PATCH] fs/select: avoid clang stack usage warning Kees Cook
2022-10-09 10:49       ` David Laight
2022-10-08  1:46   ` Andi Kleen
2022-10-11 20:50     ` Nick Desaulniers
2022-10-13 20:52       ` Andi Kleen

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=Y0nlJ6whtJuZddjr@dev-arch.thelio-3990X \
    --to=nathan@kernel.org \
    --cc=ak@linux.intel.com \
    --cc=akpm@linux-foundation.org \
    --cc=arnd@arndb.de \
    --cc=darrick.wong@oracle.com \
    --cc=edumazet@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=hch@lst.de \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=llvm@lists.linux.dev \
    --cc=mcgrof@kernel.org \
    --cc=ndesaulniers@google.com \
    --cc=nixiaoming@huawei.com \
    --cc=trix@redhat.com \
    --cc=viro@zeniv.linux.org.uk \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.