From: Oleg Nesterov <oleg@redhat.com> To: Milton Miller <miltonm@bga.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>, Andrew Morton <akpm@linux-foundation.org>, Linus Torvalds <torvalds@linux-foundation.org>, LKML <linux-kernel@vger.kernel.org>, linux-mm <linux-mm@kvack.org>, pageexec@freemail.hu, Solar Designer <solar@openwall.com>, Eugene Teo <eteo@redhat.com>, Brad Spengler <spender@grsecurity.net>, Roland McGrath <roland@redhat.com> Subject: Re: (No subject header) Date: Wed, 1 Dec 2010 19:27:47 +0100 [thread overview] Message-ID: <20101201182747.GB6143@redhat.com> (raw) In-Reply-To: <compat-not-unlikely@mdm.bga.com> On 12/01, Milton Miller wrote: > > On Tue, 30 Nov 2010 about 20:01:29 -0000, Oleg Nesterov wrote: > > Teach get_arg_ptr() to handle compat = T case correctly. > > > #include <asm/uaccess.h> > > #include <asm/mmu_context.h> > > @@ -395,6 +396,18 @@ get_arg_ptr(const char __user * const __ > > { > > const char __user *ptr; > > > > +#ifdef CONFIG_COMPAT > > + if (unlikely(compat)) { > > This should not be marked unlikely. Unlikely tells gcc the path > with over 99% confidence and disables branch predictors on some > architectures. If called from a compat processes this will result > in a mispredicted branch every iteration. Just use if (compat) > and let the hardware branch predictors do their job. This applies to almost every likely/unlikely, and I think that compat processes should fall into "unlikely category". But I don't really mind, I can remove this hint, I added it mostly as documentation. > > +#ifdef CONFIG_COMPAT > > +int compat_do_execve(char * filename, > > + compat_uptr_t __user *argv, > > + compat_uptr_t __user *envp, > > + struct pt_regs * regs) > > +{ > > + return do_execve_common(filename, > > + (void __user*)argv, (void __user*)envp, > > Shouldn't these be compat_ptr(argv)? (makes a difference on s390) I'll recheck, but I don't think so. Please note that compat_ptr() accepts "compat_uptr_t", not "compat_uptr_t *". argv should be correct as a pointer to user-space, otherwise the current code is buggy. For example, compat_do_execve() passes argv to compat_count() which does get_user(argv) without any conversion. IOW, even if this should be fixed, I think this have nothing to do with this patch. But I'll recheck, thanks. Oleg.
WARNING: multiple messages have this Message-ID (diff)
From: Oleg Nesterov <oleg@redhat.com> To: Milton Miller <miltonm@bga.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>, Andrew Morton <akpm@linux-foundation.org>, Linus Torvalds <torvalds@linux-foundation.org>, LKML <linux-kernel@vger.kernel.org>, linux-mm <linux-mm@kvack.org>, pageexec@freemail.hu, Solar Designer <solar@openwall.com>, Eugene Teo <eteo@redhat.com>, Brad Spengler <spender@grsecurity.net>, Roland McGrath <roland@redhat.com> Subject: Re: (No subject header) Date: Wed, 1 Dec 2010 19:27:47 +0100 [thread overview] Message-ID: <20101201182747.GB6143@redhat.com> (raw) In-Reply-To: <compat-not-unlikely@mdm.bga.com> On 12/01, Milton Miller wrote: > > On Tue, 30 Nov 2010 about 20:01:29 -0000, Oleg Nesterov wrote: > > Teach get_arg_ptr() to handle compat = T case correctly. > > > #include <asm/uaccess.h> > > #include <asm/mmu_context.h> > > @@ -395,6 +396,18 @@ get_arg_ptr(const char __user * const __ > > { > > const char __user *ptr; > > > > +#ifdef CONFIG_COMPAT > > + if (unlikely(compat)) { > > This should not be marked unlikely. Unlikely tells gcc the path > with over 99% confidence and disables branch predictors on some > architectures. If called from a compat processes this will result > in a mispredicted branch every iteration. Just use if (compat) > and let the hardware branch predictors do their job. This applies to almost every likely/unlikely, and I think that compat processes should fall into "unlikely category". But I don't really mind, I can remove this hint, I added it mostly as documentation. > > +#ifdef CONFIG_COMPAT > > +int compat_do_execve(char * filename, > > + compat_uptr_t __user *argv, > > + compat_uptr_t __user *envp, > > + struct pt_regs * regs) > > +{ > > + return do_execve_common(filename, > > + (void __user*)argv, (void __user*)envp, > > Shouldn't these be compat_ptr(argv)? (makes a difference on s390) I'll recheck, but I don't think so. Please note that compat_ptr() accepts "compat_uptr_t", not "compat_uptr_t *". argv should be correct as a pointer to user-space, otherwise the current code is buggy. For example, compat_do_execve() passes argv to compat_count() which does get_user(argv) without any conversion. IOW, even if this should be fixed, I think this have nothing to do with this patch. But I'll recheck, thanks. Oleg. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2010-12-01 18:34 UTC|newest] Thread overview: 231+ messages / expand[flat|nested] mbox.gz Atom feed top 2010-10-25 3:26 [resend][PATCH 1/4] oom: remove totalpage normalization from oom_badness() KOSAKI Motohiro 2010-10-25 3:26 ` KOSAKI Motohiro 2010-10-25 3:27 ` [resend][PATCH 2/4] Revert "oom: deprecate oom_adj tunable" KOSAKI Motohiro 2010-10-25 3:27 ` KOSAKI Motohiro 2010-10-25 20:40 ` David Rientjes 2010-10-25 20:40 ` David Rientjes 2010-10-26 13:01 ` KOSAKI Motohiro 2010-10-26 13:01 ` KOSAKI Motohiro 2010-10-26 19:37 ` David Rientjes 2010-10-26 19:37 ` David Rientjes 2010-11-01 7:06 ` KOSAKI Motohiro 2010-11-01 7:06 ` KOSAKI Motohiro 2010-11-01 19:36 ` David Rientjes 2010-11-01 19:36 ` David Rientjes 2010-11-09 2:26 ` KOSAKI Motohiro 2010-11-09 2:26 ` KOSAKI Motohiro 2010-11-09 3:28 ` KOSAKI Motohiro 2010-11-09 3:28 ` KOSAKI Motohiro 2010-11-15 0:24 ` KOSAKI Motohiro 2010-11-15 0:24 ` KOSAKI Motohiro 2010-11-15 9:59 ` David Rientjes 2010-11-15 9:59 ` David Rientjes 2010-11-09 23:33 ` David Rientjes 2010-11-09 23:33 ` David Rientjes 2010-11-09 23:35 ` Alan Cox 2010-11-09 23:35 ` Alan Cox 2010-11-09 23:48 ` David Rientjes 2010-11-09 23:48 ` David Rientjes 2010-11-09 23:55 ` [patch] oom: document obsolete oom_adj tunable David Rientjes 2010-11-09 23:55 ` David Rientjes 2010-11-15 0:22 ` KOSAKI Motohiro 2010-11-15 0:22 ` KOSAKI Motohiro 2010-11-15 10:38 ` David Rientjes 2010-11-15 10:38 ` David Rientjes 2010-11-23 7:16 ` KOSAKI Motohiro 2010-11-23 7:16 ` KOSAKI Motohiro 2010-11-14 5:07 ` [resend][PATCH 2/4] Revert "oom: deprecate oom_adj tunable" KOSAKI Motohiro 2010-11-14 5:07 ` KOSAKI Motohiro 2010-11-14 21:39 ` David Rientjes 2010-11-14 21:39 ` David Rientjes 2010-11-23 7:16 ` KOSAKI Motohiro 2010-11-23 7:16 ` KOSAKI Motohiro 2010-11-28 1:41 ` David Rientjes 2010-11-28 1:41 ` David Rientjes 2010-11-30 13:03 ` KOSAKI Motohiro 2010-11-30 13:03 ` KOSAKI Motohiro 2010-11-30 20:07 ` David Rientjes 2010-11-30 20:07 ` David Rientjes 2010-10-25 3:28 ` [resend][PATCH 3/4] move cred_guard_mutex from task_struct to signal_struct KOSAKI Motohiro 2010-10-25 3:28 ` KOSAKI Motohiro 2010-10-25 17:26 ` Roland McGrath 2010-10-25 17:26 ` Roland McGrath 2010-10-25 17:42 ` Oleg Nesterov 2010-10-25 17:42 ` Oleg Nesterov 2010-10-25 17:51 ` Roland McGrath 2010-10-25 17:51 ` Roland McGrath 2010-10-26 13:04 ` KOSAKI Motohiro 2010-10-26 13:04 ` KOSAKI Motohiro 2010-10-26 13:18 ` Roland McGrath 2010-10-26 13:18 ` Roland McGrath 2010-10-25 3:29 ` [resend][PATCH 4/4] oom: don't ignore rss in nascent mm KOSAKI Motohiro 2010-10-25 3:29 ` KOSAKI Motohiro 2010-10-25 11:28 ` pageexec 2010-10-25 11:28 ` pageexec 2010-10-26 7:25 ` KOSAKI Motohiro 2010-10-26 7:25 ` KOSAKI Motohiro 2010-11-23 14:34 ` Oleg Nesterov 2010-11-23 14:34 ` Oleg Nesterov 2010-11-24 0:24 ` KOSAKI Motohiro 2010-11-24 0:24 ` KOSAKI Motohiro 2010-11-24 11:09 ` Oleg Nesterov 2010-11-24 11:09 ` Oleg Nesterov 2010-11-25 11:06 ` KOSAKI Motohiro 2010-11-25 11:06 ` KOSAKI Motohiro 2010-11-25 14:02 ` Oleg Nesterov 2010-11-25 14:02 ` Oleg Nesterov 2010-11-25 19:36 ` Oleg Nesterov 2010-11-25 19:36 ` Oleg Nesterov 2010-11-29 5:25 ` KOSAKI Motohiro 2010-11-29 5:25 ` KOSAKI Motohiro 2010-11-29 11:33 ` Oleg Nesterov 2010-11-29 11:33 ` Oleg Nesterov 2010-11-29 18:23 ` Oleg Nesterov 2010-11-29 18:23 ` Oleg Nesterov 2010-11-30 19:54 ` [PATCH 0/2] exec: more excessive argument size fixes for 2.6.37/stable Oleg Nesterov 2010-11-30 19:54 ` Oleg Nesterov 2010-11-30 19:55 ` [PATCH 1/2] exec: make argv/envp memory visible to oom-killer Oleg Nesterov 2010-11-30 19:55 ` Oleg Nesterov 2010-12-01 0:12 ` KOSAKI Motohiro 2010-12-01 0:12 ` KOSAKI Motohiro 2010-12-01 18:07 ` Oleg Nesterov 2010-12-01 18:07 ` Oleg Nesterov 2010-11-30 19:56 ` [PATCH 2/2] exec: copy-and-paste the fixes into compat_do_execve() paths Oleg Nesterov 2010-11-30 19:56 ` Oleg Nesterov 2010-12-01 3:04 ` KOSAKI Motohiro 2010-12-01 3:04 ` KOSAKI Motohiro 2010-11-30 20:00 ` [PATCH 0/4] exec: unify compat/non-compat code Oleg Nesterov 2010-11-30 20:00 ` Oleg Nesterov 2010-11-30 20:00 ` [PATCH 1/4] exec: introduce get_arg_ptr() helper Oleg Nesterov 2010-11-30 20:00 ` Oleg Nesterov 2010-11-30 20:01 ` [PATCH 2/4] exec: introduce "bool compat" argument Oleg Nesterov 2010-11-30 20:01 ` Oleg Nesterov 2010-11-30 20:01 ` [PATCH 3/4] exec: unify compat_do_execve() code Oleg Nesterov 2010-11-30 20:01 ` Oleg Nesterov 2010-12-01 17:37 ` (No subject header) Milton Miller 2010-12-01 17:37 ` Milton Miller 2010-12-01 18:27 ` Oleg Nesterov [this message] 2010-12-01 18:27 ` Oleg Nesterov 2011-02-25 17:52 ` [PATCH 0/4 RESEND] exec: unify compat/non-compat code Oleg Nesterov 2011-02-25 17:52 ` Oleg Nesterov 2011-02-25 17:52 ` [PATCH 1/5] exec: introduce get_arg_ptr() helper Oleg Nesterov 2011-02-25 17:52 ` Oleg Nesterov 2011-02-25 17:52 ` [PATCH 2/5] exec: introduce "bool compat" argument Oleg Nesterov 2011-02-25 17:52 ` Oleg Nesterov 2011-02-25 18:57 ` Linus Torvalds 2011-02-25 18:57 ` Linus Torvalds 2011-02-26 12:37 ` Oleg Nesterov 2011-02-26 12:37 ` Oleg Nesterov 2011-02-25 17:53 ` [PATCH 3/5] exec: unify compat_do_execve() code Oleg Nesterov 2011-02-25 17:53 ` Oleg Nesterov 2011-02-25 19:10 ` Linus Torvalds 2011-02-25 19:10 ` Linus Torvalds 2011-02-26 12:37 ` Oleg Nesterov 2011-02-26 12:37 ` Oleg Nesterov 2011-02-26 12:57 ` Oleg Nesterov 2011-02-26 12:57 ` Oleg Nesterov 2011-02-26 15:55 ` Linus Torvalds 2011-02-26 15:55 ` Linus Torvalds 2011-02-26 17:44 ` Oleg Nesterov 2011-02-26 17:44 ` Oleg Nesterov 2011-03-01 20:47 ` [PATCH v2 0/5] exec: unify native/compat code Oleg Nesterov 2011-03-01 20:47 ` Oleg Nesterov 2011-03-01 20:48 ` [PATCH v2 1/5] exec: introduce get_arg_ptr() helper Oleg Nesterov 2011-03-01 20:48 ` Oleg Nesterov 2011-03-01 20:48 ` [PATCH v2 2/5] exec: introduce "bool compat" argument Oleg Nesterov 2011-03-01 20:48 ` Oleg Nesterov 2011-03-01 20:48 ` [PATCH v2 3/5] exec: introduce conditional_user_ptr_t Oleg Nesterov 2011-03-01 20:48 ` Oleg Nesterov 2011-03-01 20:49 ` [PATCH v2 4/5] exec: unify do_execve/compat_do_execve code Oleg Nesterov 2011-03-01 20:49 ` Oleg Nesterov 2011-03-01 20:49 ` [PATCH v2 5/5] exec: document acct_arg_size() Oleg Nesterov 2011-03-01 20:49 ` Oleg Nesterov 2011-03-01 21:39 ` [PATCH v2 0/5] exec: unify native/compat code Linus Torvalds 2011-03-01 21:39 ` Linus Torvalds 2011-03-02 16:26 ` [PATCH v3 0/4] " Oleg Nesterov 2011-03-02 16:26 ` Oleg Nesterov 2011-03-02 16:27 ` [PATCH v3 1/4] exec: introduce get_arg_ptr() helper Oleg Nesterov 2011-03-02 16:27 ` Oleg Nesterov 2011-03-03 3:01 ` KOSAKI Motohiro 2011-03-03 3:01 ` KOSAKI Motohiro 2011-03-03 15:47 ` Oleg Nesterov 2011-03-03 15:47 ` Oleg Nesterov 2011-03-03 16:07 ` Linus Torvalds 2011-03-03 16:07 ` Linus Torvalds 2011-03-05 20:30 ` [PATCH v4 0/4] exec: unify native/compat code Oleg Nesterov 2011-03-05 20:30 ` Oleg Nesterov 2011-03-05 20:31 ` [PATCH v4 1/4] exec: introduce get_user_arg_ptr() helper Oleg Nesterov 2011-03-05 20:31 ` Oleg Nesterov 2011-03-05 20:31 ` [PATCH v4 2/4] exec: introduce struct user_arg_ptr Oleg Nesterov 2011-03-05 20:31 ` Oleg Nesterov 2011-03-05 20:31 ` [PATCH v4 3/4] exec: unify do_execve/compat_do_execve code Oleg Nesterov 2011-03-05 20:31 ` Oleg Nesterov 2011-03-05 20:52 ` Linus Torvalds 2011-03-05 20:52 ` Linus Torvalds 2011-03-05 21:20 ` Oleg Nesterov 2011-03-05 21:20 ` Oleg Nesterov 2011-03-05 20:31 ` [PATCH v4 4/4] exec: document acct_arg_size() Oleg Nesterov 2011-03-05 20:31 ` Oleg Nesterov 2011-03-06 12:04 ` [PATCH v4 0/4] exec: unify native/compat code KOSAKI Motohiro 2011-03-06 12:04 ` KOSAKI Motohiro 2011-03-06 17:01 ` [PATCH v5 " Oleg Nesterov 2011-03-06 17:01 ` Oleg Nesterov 2011-03-06 17:02 ` [PATCH v5 1/4] exec: introduce get_user_arg_ptr() helper Oleg Nesterov 2011-03-06 17:02 ` Oleg Nesterov 2011-03-06 17:02 ` [PATCH v5 2/4] exec: introduce struct user_arg_ptr Oleg Nesterov 2011-03-06 17:02 ` Oleg Nesterov 2011-03-06 17:02 ` [PATCH v5 3/4] exec: unify do_execve/compat_do_execve code Oleg Nesterov 2011-03-06 17:02 ` Oleg Nesterov 2011-03-06 17:03 ` [PATCH v5 4/4] exec: document acct_arg_size() Oleg Nesterov 2011-03-06 17:03 ` Oleg Nesterov 2011-03-02 16:27 ` [PATCH v3 2/4] exec: introduce struct conditional_ptr Oleg Nesterov 2011-03-02 16:27 ` Oleg Nesterov 2011-03-03 3:08 ` KOSAKI Motohiro 2011-03-03 3:08 ` KOSAKI Motohiro 2011-03-02 16:27 ` [PATCH v3 3/4] exec: unify do_execve/compat_do_execve code Oleg Nesterov 2011-03-02 16:27 ` Oleg Nesterov 2011-03-03 3:13 ` KOSAKI Motohiro 2011-03-03 3:13 ` KOSAKI Motohiro 2011-03-02 16:28 ` [PATCH v3 4/4] exec: document acct_arg_size() Oleg Nesterov 2011-03-02 16:28 ` Oleg Nesterov 2011-03-03 3:09 ` KOSAKI Motohiro 2011-03-03 3:09 ` KOSAKI Motohiro 2011-03-02 16:44 ` [PATCH v3 0/4] exec: unify native/compat code Oleg Nesterov 2011-03-02 16:44 ` Oleg Nesterov 2011-03-02 18:00 ` Linus Torvalds 2011-03-02 18:00 ` Linus Torvalds 2011-03-02 19:40 ` David Miller 2011-03-02 19:40 ` David Miller 2011-03-02 19:48 ` Linus Torvalds 2011-03-02 19:48 ` Linus Torvalds 2011-03-02 19:54 ` David Miller 2011-03-02 19:54 ` David Miller 2011-02-25 17:53 ` [PATCH 4/5] exec: unexport acct_arg_size() and get_arg_page() Oleg Nesterov 2011-02-25 17:53 ` Oleg Nesterov 2011-02-25 17:54 ` [PATCH 5/5] exec: document acct_arg_size() Oleg Nesterov 2011-02-25 17:54 ` Oleg Nesterov 2011-02-25 18:54 ` [PATCH 0/4 RESEND] exec: unify compat/non-compat code Linus Torvalds 2011-02-25 18:54 ` Linus Torvalds 2011-02-26 12:35 ` Oleg Nesterov 2011-02-26 12:35 ` Oleg Nesterov 2010-11-30 20:01 ` [PATCH 4/4] exec: unexport acct_arg_size() and get_arg_page() Oleg Nesterov 2010-11-30 20:01 ` Oleg Nesterov 2010-12-01 3:09 ` [PATCH 0/4] exec: unify compat/non-compat code KOSAKI Motohiro 2010-12-01 3:09 ` KOSAKI Motohiro 2010-11-30 0:06 ` [resend][PATCH 4/4] oom: don't ignore rss in nascent mm KOSAKI Motohiro 2010-11-30 0:06 ` KOSAKI Motohiro 2010-10-25 20:37 ` [resend][PATCH 1/4] oom: remove totalpage normalization from oom_badness() David Rientjes 2010-10-25 20:37 ` David Rientjes -- strict thread matches above, loose matches on Subject: below -- 2014-08-26 20:52 (No subject header) gina 2010-11-09 7:27 Doubt about CRTSCTS preprocessor macro definition in termbits.h file Ramya Desai 2010-11-09 16:56 ` (No subject header) Milton Miller 2010-05-27 2:42 [PATCH 5/8] sched: clarify commment for TS_POLLING Len Brown 2010-05-27 2:42 ` idle-test patches queued for upstream Len Brown 2010-05-27 2:42 ` [PATCH 1/8] cpuidle: fail to register if !CONFIG_CPU_IDLE Len Brown 2010-05-27 5:25 ` (No subject header) Milton Miller 2010-05-27 5:25 ` Milton Miller 2010-05-27 5:47 ` Len Brown 2004-11-29 23:10 John Black 2004-11-29 23:00 John Black 2004-11-30 7:09 ` Brent Clark 2004-07-30 3:33 Jason Opperisano 2004-07-30 4:00 ` John Black 2004-07-28 16:01 John Black 2004-07-28 21:52 ` Antony Stone 2004-07-30 3:09 ` John Black
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=20101201182747.GB6143@redhat.com \ --to=oleg@redhat.com \ --cc=akpm@linux-foundation.org \ --cc=eteo@redhat.com \ --cc=kosaki.motohiro@jp.fujitsu.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=miltonm@bga.com \ --cc=pageexec@freemail.hu \ --cc=roland@redhat.com \ --cc=solar@openwall.com \ --cc=spender@grsecurity.net \ --cc=torvalds@linux-foundation.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: linkBe 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.