From: Aaron Conole <aconole@redhat.com> To: Dmitry Vyukov <dvyukov@google.com> Cc: Eric Dumazet <edumazet@google.com>, syzkaller <syzkaller@googlegroups.com>, Vladislav Yasevich <vyasevich@gmail.com>, linux-sctp@vger.kernel.org, netdev <netdev@vger.kernel.org>, Kostya Serebryany <kcc@google.com>, Alexander Potapenko <glider@google.com>, Sasha Levin <sasha.levin@oracle.com>, Joe Perches <joe@perches.com> Subject: Re: use-after-free in sctp_do_sm Date: Thu, 03 Dec 2015 13:52:12 -0500 [thread overview] Message-ID: <f7twpsvgyar.fsf@aconole.bos.csb> (raw) In-Reply-To: <CACT4Y+YJe-OJj9q1Kr5-_ApDt1UBKLjTRxhP62EHtEAwZt5vXw@mail.gmail.com> (Dmitry Vyukov's message of "Thu, 3 Dec 2015 18:12:40 +0100") Dmitry Vyukov <dvyukov@google.com> writes: > On Thu, Dec 3, 2015 at 6:02 PM, Eric Dumazet <edumazet@google.com> wrote: >> On Thu, Dec 3, 2015 at 7:55 AM, Dmitry Vyukov <dvyukov@google.com> wrote: >>> On Thu, Dec 3, 2015 at 3:48 PM, Eric Dumazet <edumazet@google.com> wrote: >>>>> >>>>> No, I don't. But pr_debug always computes its arguments. See no_printk >>>>> in printk.h. So this use-after-free happens for all users. >>>> >>>> Hmm. >>>> >>>> pr_debug() should be a nop unless either DEBUG or >>>> CONFIG_DYNAMIC_DEBUG are set >>>> >>>> On our production kernels, pr_debug() is a nop. >>>> >>>> Can you double check ? Thanks ! >>> >>> >>> Why should it be nop? no_printk thing in printk.h pretty much >>> explicitly makes it not a nop... Because it was until commit 5264f2f75d8. It also violates my reading of the following from printk.h: * All of these will print unconditionally, although note that pr_debug() * and other debug macros are compiled out unless either DEBUG is defined * or CONFIG_DYNAMIC_DEBUG is set. >>> >>> Double-checked: debug_post_sfx leads to some generated code: >>> >>> debug_post_sfx(); >>> ffffffff8229f256: 48 8b 85 58 fe ff ff mov -0x1a8(%rbp),%rax >>> ffffffff8229f25d: 48 85 c0 test %rax,%rax >>> ffffffff8229f260: 74 24 je >>> ffffffff8229f286 <sctp_do_sm+0x176> >>> ffffffff8229f262: 8b b0 a8 00 00 00 mov 0xa8(%rax),%esi >>> ffffffff8229f268: 48 8b 85 60 fe ff ff mov -0x1a0(%rbp),%rax >>> ffffffff8229f26f: 44 89 85 74 fe ff ff mov %r8d,-0x18c(%rbp) >>> ffffffff8229f276: 48 8b 78 20 mov 0x20(%rax),%rdi >>> ffffffff8229f27a: e8 71 28 01 00 callq >>> ffffffff822b1af0 <sctp_id2assoc> >>> ffffffff8229f27f: 44 8b 85 74 fe ff ff mov -0x18c(%rbp),%r8d >>> >>> return error; >>> } >>> ffffffff8229f286: 48 81 c4 a0 01 00 00 add $0x1a0,%rsp >>> ffffffff8229f28d: 44 89 c0 mov %r8d,%eax >>> ffffffff8229f290: 5b pop %rbx >>> ffffffff8229f291: 41 5c pop %r12 >>> ffffffff8229f293: 41 5d pop %r13 >>> ffffffff8229f295: 41 5e pop %r14 >>> ffffffff8229f297: 41 5f pop %r15 >>> ffffffff8229f299: 5d pop %rbp >>> ffffffff8229f29a: c3 retq >> >> This is a serious concern, because we let in the past lot of patches >> converting traditional +1 >> #ifdef DEBUG >> # define some_hand_coded_ugly_debug() printk( ...._ >> #else >> # define some_hand_coded_ugly_debug() >> #endif >> >> On the premise pr_debug() would be a nop. >> >> It seems it is not always the case. This is a very serious problem. +1 >> We probably have hundred of potential bugs, because few people >> actually make sure all debugging stuff is correct, >> like comments can be wrong because they are not updated properly as >> time flies. >> >> It is definitely a nop for many cases. >> >> +void eric_test_pr_debug(struct sock *sk) >> +{ >> + if (atomic_read(&sk->sk_omem_alloc)) >> + pr_debug("%s: optmem leakage for sock %p\n", >> + __func__, sk); >> +} >> >> -> >> >> 0000000000004740 <eric_test_pr_debug>: >> 4740: e8 00 00 00 00 callq 4745 <eric_test_pr_debug+0x5> >> 4741: R_X86_64_PC32 __fentry__-0x4 >> 4745: 55 push %rbp >> 4746: 8b 87 24 01 00 00 mov 0x124(%rdi),%eax // >> atomic_read() but nothing follows >> 474c: 48 89 e5 mov %rsp,%rbp >> 474f: 5d pop %rbp >> 4750: c3 retq > > > > I would expect that it is nop when argument evaluation does not have > side-effects. For example, for a load of a variable compiler will most > likely elide it (though, it does not have to elide it, because the > load is spelled in the code, so it can also legally emit the load and > doesn't use the result). > But if argument computation has side-effect (or compiler can't prove > otherwise), it must emit code. It must emit code for function calls > when the function is defined in a different translation unit, and for > volatile accesses (most likely including atomic accesses), etc This isn't 100% true. As you state, in order to reach the return 0, all side effects must be evaluated. Load generally does not have side effects, so it can be safely elided, but function() must be emitted. However, that is _not_ required to get the desired warning emission on a printf argument function, see http://pastebin.com/UHuaydkj for an example. I think that as a minimum, the following patch should be evaluted, but am unsure to whom I should submit it (after I test): diff --git a/include/linux/printk.h b/include/linux/printk.h index 9729565..cd24d2d 100644 --- a/include/linux/printk.h +++ b/include/linux/printk.h @@ -286,7 +286,7 @@ extern asmlinkage void dump_stack(void) __cold; printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__) #else #define pr_debug(fmt, ...) \ - no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__) + ({ if(0) printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__); 0;}) #endif /*
WARNING: multiple messages have this Message-ID (diff)
From: Aaron Conole <aconole@redhat.com> To: Dmitry Vyukov <dvyukov@google.com> Cc: Eric Dumazet <edumazet@google.com>, syzkaller <syzkaller@googlegroups.com>, Vladislav Yasevich <vyasevich@gmail.com>, linux-sctp@vger.kernel.org, netdev <netdev@vger.kernel.org>, Kostya Serebryany <kcc@google.com>, Alexander Potapenko <glider@google.com>, Sasha Levin <sasha.levin@oracle.com>, Joe Perches <joe@perches.com> Subject: Re: use-after-free in sctp_do_sm Date: Thu, 03 Dec 2015 18:52:12 +0000 [thread overview] Message-ID: <f7twpsvgyar.fsf@aconole.bos.csb> (raw) In-Reply-To: <CACT4Y+YJe-OJj9q1Kr5-_ApDt1UBKLjTRxhP62EHtEAwZt5vXw@mail.gmail.com> (Dmitry Vyukov's message of "Thu, 3 Dec 2015 18:12:40 +0100") Dmitry Vyukov <dvyukov@google.com> writes: > On Thu, Dec 3, 2015 at 6:02 PM, Eric Dumazet <edumazet@google.com> wrote: >> On Thu, Dec 3, 2015 at 7:55 AM, Dmitry Vyukov <dvyukov@google.com> wrote: >>> On Thu, Dec 3, 2015 at 3:48 PM, Eric Dumazet <edumazet@google.com> wrote: >>>>> >>>>> No, I don't. But pr_debug always computes its arguments. See no_printk >>>>> in printk.h. So this use-after-free happens for all users. >>>> >>>> Hmm. >>>> >>>> pr_debug() should be a nop unless either DEBUG or >>>> CONFIG_DYNAMIC_DEBUG are set >>>> >>>> On our production kernels, pr_debug() is a nop. >>>> >>>> Can you double check ? Thanks ! >>> >>> >>> Why should it be nop? no_printk thing in printk.h pretty much >>> explicitly makes it not a nop... Because it was until commit 5264f2f75d8. It also violates my reading of the following from printk.h: * All of these will print unconditionally, although note that pr_debug() * and other debug macros are compiled out unless either DEBUG is defined * or CONFIG_DYNAMIC_DEBUG is set. >>> >>> Double-checked: debug_post_sfx leads to some generated code: >>> >>> debug_post_sfx(); >>> ffffffff8229f256: 48 8b 85 58 fe ff ff mov -0x1a8(%rbp),%rax >>> ffffffff8229f25d: 48 85 c0 test %rax,%rax >>> ffffffff8229f260: 74 24 je >>> ffffffff8229f286 <sctp_do_sm+0x176> >>> ffffffff8229f262: 8b b0 a8 00 00 00 mov 0xa8(%rax),%esi >>> ffffffff8229f268: 48 8b 85 60 fe ff ff mov -0x1a0(%rbp),%rax >>> ffffffff8229f26f: 44 89 85 74 fe ff ff mov %r8d,-0x18c(%rbp) >>> ffffffff8229f276: 48 8b 78 20 mov 0x20(%rax),%rdi >>> ffffffff8229f27a: e8 71 28 01 00 callq >>> ffffffff822b1af0 <sctp_id2assoc> >>> ffffffff8229f27f: 44 8b 85 74 fe ff ff mov -0x18c(%rbp),%r8d >>> >>> return error; >>> } >>> ffffffff8229f286: 48 81 c4 a0 01 00 00 add $0x1a0,%rsp >>> ffffffff8229f28d: 44 89 c0 mov %r8d,%eax >>> ffffffff8229f290: 5b pop %rbx >>> ffffffff8229f291: 41 5c pop %r12 >>> ffffffff8229f293: 41 5d pop %r13 >>> ffffffff8229f295: 41 5e pop %r14 >>> ffffffff8229f297: 41 5f pop %r15 >>> ffffffff8229f299: 5d pop %rbp >>> ffffffff8229f29a: c3 retq >> >> This is a serious concern, because we let in the past lot of patches >> converting traditional +1 >> #ifdef DEBUG >> # define some_hand_coded_ugly_debug() printk( ...._ >> #else >> # define some_hand_coded_ugly_debug() >> #endif >> >> On the premise pr_debug() would be a nop. >> >> It seems it is not always the case. This is a very serious problem. +1 >> We probably have hundred of potential bugs, because few people >> actually make sure all debugging stuff is correct, >> like comments can be wrong because they are not updated properly as >> time flies. >> >> It is definitely a nop for many cases. >> >> +void eric_test_pr_debug(struct sock *sk) >> +{ >> + if (atomic_read(&sk->sk_omem_alloc)) >> + pr_debug("%s: optmem leakage for sock %p\n", >> + __func__, sk); >> +} >> >> -> >> >> 0000000000004740 <eric_test_pr_debug>: >> 4740: e8 00 00 00 00 callq 4745 <eric_test_pr_debug+0x5> >> 4741: R_X86_64_PC32 __fentry__-0x4 >> 4745: 55 push %rbp >> 4746: 8b 87 24 01 00 00 mov 0x124(%rdi),%eax // >> atomic_read() but nothing follows >> 474c: 48 89 e5 mov %rsp,%rbp >> 474f: 5d pop %rbp >> 4750: c3 retq > > > > I would expect that it is nop when argument evaluation does not have > side-effects. For example, for a load of a variable compiler will most > likely elide it (though, it does not have to elide it, because the > load is spelled in the code, so it can also legally emit the load and > doesn't use the result). > But if argument computation has side-effect (or compiler can't prove > otherwise), it must emit code. It must emit code for function calls > when the function is defined in a different translation unit, and for > volatile accesses (most likely including atomic accesses), etc This isn't 100% true. As you state, in order to reach the return 0, all side effects must be evaluated. Load generally does not have side effects, so it can be safely elided, but function() must be emitted. However, that is _not_ required to get the desired warning emission on a printf argument function, see http://pastebin.com/UHuaydkj for an example. I think that as a minimum, the following patch should be evaluted, but am unsure to whom I should submit it (after I test): diff --git a/include/linux/printk.h b/include/linux/printk.h index 9729565..cd24d2d 100644 --- a/include/linux/printk.h +++ b/include/linux/printk.h @@ -286,7 +286,7 @@ extern asmlinkage void dump_stack(void) __cold; printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__) #else #define pr_debug(fmt, ...) \ - no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__) + ({ if(0) printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__); 0;}) #endif /*
next prev parent reply other threads:[~2015-12-03 18:52 UTC|newest] Thread overview: 153+ messages / expand[flat|nested] mbox.gz Atom feed top 2015-11-24 9:15 use-after-free in sctp_do_sm Dmitry Vyukov 2015-11-24 9:15 ` Dmitry Vyukov 2015-11-24 9:31 ` Dmitry Vyukov 2015-11-24 9:31 ` Dmitry Vyukov 2015-11-24 10:10 ` Dmitry Vyukov 2015-11-24 10:10 ` Dmitry Vyukov 2015-11-24 20:45 ` Neil Horman 2015-11-24 20:45 ` Neil Horman 2015-11-24 21:08 ` Eric Dumazet 2015-11-24 21:08 ` Eric Dumazet 2015-11-24 21:12 ` David Miller 2015-11-24 21:12 ` David Miller 2015-11-25 15:12 ` Vlad Yasevich 2015-11-25 15:12 ` Vlad Yasevich 2015-11-28 15:50 ` Dmitry Vyukov 2015-11-28 15:50 ` Dmitry Vyukov 2015-12-03 16:51 ` Marcelo Ricardo Leitner 2015-12-03 16:51 ` Marcelo Ricardo Leitner 2015-12-03 17:43 ` Marcelo Ricardo Leitner 2015-12-03 17:43 ` Marcelo Ricardo Leitner 2015-12-03 17:59 ` Eric Dumazet 2015-12-03 17:59 ` Eric Dumazet 2015-12-03 18:06 ` Marcelo 2015-12-03 18:06 ` Marcelo 2015-12-03 18:35 ` Vlad Yasevich 2015-12-03 18:35 ` Vlad Yasevich 2015-12-03 18:43 ` Marcelo 2015-12-03 18:43 ` Marcelo 2015-12-04 17:14 ` [PATCH net 0/3] sctp: packet timestamp fixes Marcelo Ricardo Leitner 2015-12-04 17:14 ` Marcelo Ricardo Leitner 2015-12-04 17:14 ` [PATCH net 1/3] sctp: use the same clock as if sock source timestamps were on Marcelo Ricardo Leitner 2015-12-04 17:14 ` Marcelo Ricardo Leitner 2015-12-04 20:31 ` Vlad Yasevich 2015-12-04 20:31 ` Vlad Yasevich 2015-12-04 17:14 ` [PATCH net 2/3] sctp: update the netstamp_needed counter when copying sockets Marcelo Ricardo Leitner 2015-12-04 17:14 ` Marcelo Ricardo Leitner 2015-12-04 20:33 ` Vlad Yasevich 2015-12-04 20:33 ` Vlad Yasevich 2015-12-04 17:14 ` [PATCH net 3/3] sctp: also copy sk_tsflags when copying the socket Marcelo Ricardo Leitner 2015-12-04 17:14 ` Marcelo Ricardo Leitner 2015-12-04 20:33 ` Vlad Yasevich 2015-12-04 20:33 ` Vlad Yasevich 2015-12-06 3:24 ` [PATCH net 0/3] sctp: packet timestamp fixes David Miller 2015-12-06 3:24 ` David Miller 2015-12-03 13:05 ` use-after-free in sctp_do_sm Marcelo Ricardo Leitner 2015-12-03 13:05 ` Marcelo Ricardo Leitner 2015-12-03 13:45 ` Dmitry Vyukov 2015-12-03 13:45 ` Dmitry Vyukov 2015-12-03 14:48 ` Eric Dumazet 2015-12-03 14:48 ` Eric Dumazet 2015-12-03 15:55 ` Dmitry Vyukov 2015-12-03 15:55 ` Dmitry Vyukov 2015-12-03 16:15 ` Marcelo Ricardo Leitner 2015-12-03 16:15 ` Marcelo Ricardo Leitner 2015-12-03 17:02 ` Eric Dumazet 2015-12-03 17:02 ` Eric Dumazet 2015-12-03 17:12 ` Dmitry Vyukov 2015-12-03 17:12 ` Dmitry Vyukov 2015-12-03 18:52 ` Aaron Conole [this message] 2015-12-03 18:52 ` Aaron Conole 2015-12-03 19:06 ` Joe Perches 2015-12-03 19:06 ` Joe Perches 2015-12-03 19:32 ` Jason Baron 2015-12-03 19:32 ` Jason Baron 2015-12-03 20:03 ` Joe Perches 2015-12-03 20:03 ` Joe Perches 2015-12-03 20:10 ` Jason Baron 2015-12-03 20:10 ` Jason Baron 2015-12-03 20:24 ` Joe Perches 2015-12-03 20:24 ` Joe Perches 2015-12-03 20:42 ` Jason Baron 2015-12-03 20:42 ` Jason Baron 2015-12-03 20:51 ` Joe Perches 2015-12-03 20:51 ` Joe Perches 2015-12-04 10:40 ` Dmitry Vyukov 2015-12-04 10:40 ` Dmitry Vyukov 2015-12-04 12:55 ` Marcelo Ricardo Leitner 2015-12-04 12:55 ` Marcelo Ricardo Leitner 2015-12-04 15:37 ` Vlad Yasevich 2015-12-04 15:37 ` Vlad Yasevich 2015-12-04 15:51 ` Aaron Conole 2015-12-04 15:51 ` Aaron Conole 2015-12-04 16:12 ` Dmitry Vyukov 2015-12-04 16:12 ` Dmitry Vyukov 2015-12-04 16:47 ` Jason Baron 2015-12-04 16:47 ` Jason Baron 2015-12-04 17:03 ` Joe Perches 2015-12-04 17:03 ` Joe Perches 2015-12-04 17:11 ` Jason Baron 2015-12-04 17:11 ` Jason Baron 2015-12-04 10:41 ` Dmitry Vyukov 2015-12-04 10:41 ` Dmitry Vyukov 2015-12-04 17:48 ` Marcelo Ricardo Leitner 2015-12-04 17:48 ` Marcelo Ricardo Leitner 2015-12-04 20:25 ` Dmitry Vyukov 2015-12-04 20:25 ` Dmitry Vyukov 2015-12-04 21:34 ` Marcelo Ricardo Leitner 2015-12-04 21:34 ` Marcelo Ricardo Leitner 2015-12-04 21:38 ` Dmitry Vyukov 2015-12-04 21:38 ` Dmitry Vyukov 2015-12-05 16:39 ` Vlad Yasevich 2015-12-05 16:39 ` Vlad Yasevich 2015-12-07 11:26 ` Dmitry Vyukov 2015-12-07 11:26 ` Dmitry Vyukov 2015-12-07 13:15 ` Marcelo Ricardo Leitner 2015-12-07 13:15 ` Marcelo Ricardo Leitner 2015-12-07 13:20 ` Dmitry Vyukov 2015-12-07 13:20 ` Dmitry Vyukov 2015-12-07 18:52 ` Marcelo Ricardo Leitner 2015-12-07 18:52 ` Marcelo Ricardo Leitner 2015-12-07 19:33 ` Vlad Yasevich 2015-12-07 19:33 ` Vlad Yasevich 2015-12-07 19:50 ` Marcelo Ricardo Leitner 2015-12-07 19:50 ` Marcelo Ricardo Leitner 2015-12-07 20:37 ` Vlad Yasevich 2015-12-07 20:37 ` Vlad Yasevich 2015-12-07 20:52 ` Marcelo Ricardo Leitner 2015-12-07 20:52 ` Marcelo Ricardo Leitner 2015-12-08 17:30 ` Dmitry Vyukov 2015-12-08 17:30 ` Dmitry Vyukov 2015-12-08 17:40 ` Marcelo Ricardo Leitner 2015-12-08 17:40 ` Marcelo Ricardo Leitner 2015-12-08 19:22 ` Dmitry Vyukov 2015-12-08 19:22 ` Dmitry Vyukov 2015-12-09 14:41 ` Dmitry Vyukov 2015-12-09 14:41 ` Dmitry Vyukov 2015-12-09 15:03 ` Marcelo Ricardo Leitner 2015-12-09 15:03 ` Marcelo Ricardo Leitner 2015-12-09 16:41 ` Marcelo Ricardo Leitner 2015-12-09 16:41 ` Marcelo Ricardo Leitner 2015-12-11 13:35 ` Dmitry Vyukov 2015-12-11 13:35 ` Dmitry Vyukov 2015-12-11 13:51 ` Marcelo Ricardo Leitner 2015-12-11 13:51 ` Marcelo Ricardo Leitner 2015-12-11 14:03 ` Marcelo Ricardo Leitner 2015-12-11 14:03 ` Marcelo Ricardo Leitner 2015-12-11 14:30 ` Dmitry Vyukov 2015-12-11 14:30 ` Dmitry Vyukov 2015-12-11 15:55 ` Marcelo Ricardo Leitner 2015-12-11 15:55 ` Marcelo Ricardo Leitner 2016-01-08 13:00 ` [PATCH] sctp: fix use-after-free in pr_debug statement Marcelo Ricardo Leitner 2016-01-08 13:00 ` Marcelo Ricardo Leitner 2016-01-11 17:00 ` Vlad Yasevich 2016-01-11 17:00 ` Vlad Yasevich 2016-01-11 22:13 ` David Miller 2016-01-11 22:13 ` David Miller 2016-01-12 8:41 ` Dmitry Vyukov 2016-01-12 8:41 ` Dmitry Vyukov 2015-12-11 18:37 ` use-after-free in sctp_do_sm Vlad Yasevich 2015-12-11 18:37 ` Vlad Yasevich 2015-12-14 9:50 ` David Laight 2015-12-14 14:25 ` Vlad Yasevich 2015-12-14 14:25 ` Vlad Yasevich
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=f7twpsvgyar.fsf@aconole.bos.csb \ --to=aconole@redhat.com \ --cc=dvyukov@google.com \ --cc=edumazet@google.com \ --cc=glider@google.com \ --cc=joe@perches.com \ --cc=kcc@google.com \ --cc=linux-sctp@vger.kernel.org \ --cc=netdev@vger.kernel.org \ --cc=sasha.levin@oracle.com \ --cc=syzkaller@googlegroups.com \ --cc=vyasevich@gmail.com \ /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.