linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Indan Zupancic" <indan@nul.nu>
To: "Eric Dumazet" <eric.dumazet@gmail.com>
Cc: "Will Drewry" <wad@chromium.org>,
	linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org,
	linux-doc@vger.kernel.org, kernel-hardening@lists.openwall.com,
	netdev@vger.kernel.org, x86@kernel.org, arnd@arndb.de,
	davem@davemloft.net, hpa@zytor.com, mingo@redhat.com,
	oleg@redhat.com, peterz@infradead.org, rdunlap@xenotime.net,
	mcgrathr@chromium.org, tglx@linutronix.de, luto@mit.edu,
	eparis@redhat.com, serge.hallyn@canonical.com, djm@mindrot.org,
	scarybeasts@gmail.com, pmoore@redhat.com,
	akpm@linux-foundation.org, corbet@lwn.net, markus@chromium.org,
	coreyb@linux.vnet.ibm.com, keescook@chromium.org
Subject: Re: [PATCH] net: bpf_jit: Simplify code by always using offset8 or offset32.
Date: Tue, 20 Mar 2012 22:33:35 +1100	[thread overview]
Message-ID: <cfb04c3fe59e421093f035c035c88b51.squirrel@webmail.greenhost.nl> (raw)
In-Reply-To: <1332212386.22737.20.camel@edumazet-glaptop>

On Tue, March 20, 2012 13:59, Eric Dumazet wrote:
> On Tue, 2012-03-20 at 13:24 +1100, Indan Zupancic wrote:
>
>> If it does then perhaps the fast path should be made faster by inlining
>> the code instead of calling a function which may not be cached.
>>
>
> inlining 400 times a sequence of code is waste of icache, you probably
> missed this.

Well, according to you most filters were small, inling 26 bytes a few
times should be faster than calling an external function. Not all calls
need to be inlined either.

>
> I spent a lot of time on working on this implementation, tried many
> different strategies before choosing the one in place.
>
> Listen, I am tired of this thread, it seems you want to push changes
> that have almost no value but still need lot of review.

The latest patch didn't change generated code except for a few ancillary
instructions. The one before that just added documentation. The first
patch was indeed bad.

>
> Unless you make benchmarks and can make at least 5 % improvement of the
> speed, or improve maintainability of this code, I am not interested.

My next patch would have changed the compiler to always compile in two
passes instead of looping till the result is stable. But never mind.

>
> We certainly _can_ one day have sizeof(struct sk_buff) > 256, and actual
> code is ready for this. You want to break this for absolutely no valid
> reason.

I've the feeling you didn't read the latest patch, it doesn't assume
sizeof(struct sk_buff) < 256, nor that fields aren't reordered.

>
> We _can_ change fields order anytime in struct sk_buff, even if you
> state "its very unlikely that those fields are ever moved to the end
> of sk_buff".

And if the dev, len or data_len fields are really moved past the first
127 bytes the JIT code can be changed too. The JIT code already depends
on some of struct sk_buff's field properties anyway.

Greetings,

Indan



  reply	other threads:[~2012-03-20 11:33 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-03-12 21:28 [PATCH v14 01/13] sk_run_filter: add BPF_S_ANC_SECCOMP_LD_W Will Drewry
2012-03-12 21:28 ` [PATCH v14 02/13] net/compat.c,linux/filter.h: share compat_sock_fprog Will Drewry
2012-03-12 21:28 ` [PATCH v14 03/13] seccomp: kill the seccomp_t typedef Will Drewry
2012-03-12 21:28 ` [PATCH v14 04/13] arch/x86: add syscall_get_arch to syscall.h Will Drewry
2012-03-12 21:28 ` [PATCH v14 05/13] asm/syscall.h: add syscall_get_arch Will Drewry
2012-03-12 21:28 ` [PATCH v14 06/13] seccomp: add system call filtering using BPF Will Drewry
2012-03-13  3:33   ` Indan Zupancic
2012-03-13 15:57     ` [kernel-hardening] " Will Drewry
2012-03-12 21:28 ` [PATCH v14 07/13] signal, x86: add SIGSYS info and make it synchronous Will Drewry
2012-03-12 21:28 ` [PATCH v14 08/13] seccomp: add SECCOMP_RET_ERRNO Will Drewry
2012-03-12 21:28 ` [PATCH v14 09/13] seccomp: Add SECCOMP_RET_TRAP Will Drewry
2012-03-12 21:28 ` [PATCH v14 10/13] ptrace,seccomp: Add PTRACE_SECCOMP support Will Drewry
2012-03-14  7:31   ` Indan Zupancic
2012-03-14 15:03     ` [kernel-hardening] " Will Drewry
2012-03-14 15:52       ` Will Drewry
2012-03-15 20:31     ` [PATCH v16 11/13] " Will Drewry
2012-03-12 21:28 ` [PATCH v14 11/13] x86: Enable HAVE_ARCH_SECCOMP_FILTER Will Drewry
2012-03-12 21:28 ` [PATCH v14 12/13] Documentation: prctl/seccomp_filter Will Drewry
2012-03-12 21:28 ` [PATCH v14 13/13] seccomp: remove duplicated failure logging Will Drewry
2012-03-13  3:40 ` [PATCH v14 01/13] sk_run_filter: add BPF_S_ANC_SECCOMP_LD_W Indan Zupancic
2012-03-13 15:40   ` Will Drewry
2012-03-13 10:04 ` Indan Zupancic
2012-03-13 15:43   ` Will Drewry
2012-03-13 17:13   ` Eric Dumazet
2012-03-14  5:12     ` Indan Zupancic
2012-03-14  5:55       ` Eric Dumazet
2012-03-14  7:59         ` Indan Zupancic
2012-03-14  8:05           ` Eric Dumazet
2012-03-17 10:14             ` Indan Zupancic
2012-03-17 13:49               ` Eric Dumazet
2012-03-18  8:35                 ` Indan Zupancic
2012-03-18 12:40                   ` [PATCH] net: bpf_jit: fix BPF_S_LDX_B_MSH compilation Eric Dumazet
2012-03-19 21:42                     ` David Miller
2012-03-20  0:16                     ` [PATCH] net: bpf_jit: Document evilness of negative indirect loads Indan Zupancic
2012-03-18 12:52                   ` [PATCH v14 01/13] sk_run_filter: add BPF_S_ANC_SECCOMP_LD_W Eric Dumazet
2012-03-20  2:24                     ` [PATCH] net: bpf_jit: Simplify code by always using offset8 or offset32 Indan Zupancic
2012-03-20  2:59                       ` Eric Dumazet
2012-03-20 11:33                         ` Indan Zupancic [this message]
2012-03-20 11:41                           ` David Laight
2012-03-20 13:56                           ` Eric Dumazet

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=cfb04c3fe59e421093f035c035c88b51.squirrel@webmail.greenhost.nl \
    --to=indan@nul.nu \
    --cc=akpm@linux-foundation.org \
    --cc=arnd@arndb.de \
    --cc=corbet@lwn.net \
    --cc=coreyb@linux.vnet.ibm.com \
    --cc=davem@davemloft.net \
    --cc=djm@mindrot.org \
    --cc=eparis@redhat.com \
    --cc=eric.dumazet@gmail.com \
    --cc=hpa@zytor.com \
    --cc=keescook@chromium.org \
    --cc=kernel-hardening@lists.openwall.com \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=luto@mit.edu \
    --cc=markus@chromium.org \
    --cc=mcgrathr@chromium.org \
    --cc=mingo@redhat.com \
    --cc=netdev@vger.kernel.org \
    --cc=oleg@redhat.com \
    --cc=peterz@infradead.org \
    --cc=pmoore@redhat.com \
    --cc=rdunlap@xenotime.net \
    --cc=scarybeasts@gmail.com \
    --cc=serge.hallyn@canonical.com \
    --cc=tglx@linutronix.de \
    --cc=wad@chromium.org \
    --cc=x86@kernel.org \
    /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 NNTP newsgroup(s).