| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <BANLkTikYwWe6E6mb3PLE3L7a3VHWdtEvfQ@mail.gmail.com>
Date: Fri, 22 Apr 2011 15:19:15 +0800
From: Yong Zhang <yong.zhang0@...il.com>
To: Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>
Cc: a.p.zijlstra@...llo.nl, rostedt@...dmis.org, tglx@...utronix.de,
mingo@...e.hu, linux-kernel@...r.kernel.org
Subject: Re: [RFC][PATCH 0/7] lockdep: Support recurise-read locks
2011/4/18 Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>:
> (Continued from https://lkml.org/lkml/2011/3/31/240 "Re: [RFC] seqlock,lockdep: Add lock primitives to read_seqbegin().")
> Test results for above program:
>
> "cat /proc/locktest1 /proc/locktest2" => Detect fail
> "cat /proc/locktest2 /proc/locktest1" => Detect fail
> "cat /proc/locktest3 /proc/locktest4" => Detect fail
> "cat /proc/locktest4 /proc/locktest3" => Detect fail
>
> If I change from rwlock_acquire_read() to spin_acquire() in read_seqbegin2()
> and from rwlock_release() to spin_release() in read_seqretry2():
>
> "cat /proc/locktest1 /proc/locktest2" => Detect fail
> "cat /proc/locktest2 /proc/locktest1" => Detect OK (shown below)
> "cat /proc/locktest3 /proc/locktest4" => Detect fail
> "cat /proc/locktest4 /proc/locktest3" => Detect OK (shown below)
>
> Guessing from my testcases, read_seqbegin2() will fail to detect the problem
> unless we use spin_acquire()/spin_release() rather than
> rwlock_acquire_read()/rwlock_release().
>
> Also, something is still wrong because lockdep fails to detect the problem
> for "cat /proc/locktest1 /proc/locktest2" and
> "cat /proc/locktest3 /proc/locktest4" cases.
It fails because we never add the recursive read to prev->after evev if
it passed the validation.
check_prev_add()::1671
/*
* For recursive read-locks we do all the dependency checks,
* but we dont store read-triggered dependencies (only
* write-triggered dependencies). This ensures that only the
* write-side dependencies matter, and that if for example a
* write-lock never takes any other locks, then the reads are
* equivalent to a NOP.
*/
if (next->read == 2 || prev->read == 2)
return 1;
So we have no chain after opening locktest1 or locktest3.
But adding recursive read to prev->after will introduce spurious
lockdep warnings.
Thanks,
Yong
>
> [ 83.551455]
> [ 83.551457] =======================================================
> [ 83.555293] [ INFO: possible circular locking dependency detected ]
> [ 83.555293] 2.6.39-rc3-00228-gd733ed6-dirty #259
> [ 83.555293] -------------------------------------------------------
> [ 83.555293] cat/2768 is trying to acquire lock:
> [ 83.555293] (brlock1_lock_dep_map){++++..}, at: [<e08150b0>] brlock1_local_lock+0x0/0x90 [locktest]
> [ 83.555293]
> [ 83.555293] but task is already holding lock:
> [ 83.555293] (&(&(&seqlock1)->lock)->rlock){+.+...}, at: [<e08154dd>] locktest_open1+0xd/0x40 [locktest]
> [ 83.555293]
> [ 83.555293] which lock already depends on the new lock.
> [ 83.555293]
> [ 83.555293]
> [ 83.555293] the existing dependency chain (in reverse order) is:
> [ 83.555293]
> [ 83.555293] -> #1 (&(&(&seqlock1)->lock)->rlock){+.+...}:
> [ 83.635281] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 83.635281] [<c106c840>] validate_chain+0x320/0x5a0
> [ 83.635281] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 83.635281] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 83.635281] [<e0815555>] locktest_open2+0x45/0x70 [locktest]
> [ 83.635281] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 83.635281] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 83.635281] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 83.635281] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 83.635281] [<c10dc846>] path_openat+0xa6/0x340
> [ 83.635281] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 83.635281] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 83.635281] [<c10cf069>] sys_open+0x29/0x40
> [ 83.635281] [<c13b43c1>] syscall_call+0x7/0xb
> [ 83.635281]
> [ 83.635281] -> #0 (brlock1_lock_dep_map){++++..}:
> [ 83.635281] [<c106c34c>] check_prev_add+0x78c/0x820
> [ 83.635281] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 83.635281] [<c106c840>] validate_chain+0x320/0x5a0
> [ 83.635281] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 83.635281] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 83.635281] [<e08150e3>] brlock1_local_lock+0x33/0x90 [locktest]
> [ 83.635281] [<e08154e8>] locktest_open1+0x18/0x40 [locktest]
> [ 83.635281] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 83.635281] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 83.635281] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 83.635281] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 83.635281] [<c10dc846>] path_openat+0xa6/0x340
> [ 83.635281] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 83.635281] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 83.635281] [<c10cf069>] sys_open+0x29/0x40
> [ 83.635281] [<c13b43c1>] syscall_call+0x7/0xb
> [ 83.635281]
> [ 83.635281] other info that might help us debug this:
> [ 83.635281]
> [ 83.635281] 1 lock held by cat/2768:
> [ 83.635281] #0: (&(&(&seqlock1)->lock)->rlock){+.+...}, at: [<e08154dd>] locktest_open1+0xd/0x40 [locktest]
> [ 83.635281]
> [ 83.635281] stack backtrace:
> [ 83.635281] Pid: 2768, comm: cat Not tainted 2.6.39-rc3-00228-gd733ed6-dirty #259
> [ 83.635281] Call Trace:
> [ 83.635281] [<c106ade6>] print_circular_bug+0xc6/0xd0
> [ 83.635281] [<c106c34c>] check_prev_add+0x78c/0x820
> [ 83.635281] [<c1005d3b>] ? print_context_stack+0x3b/0xa0
> [ 83.635281] [<c1004fa1>] ? dump_trace+0x81/0xe0
> [ 83.635281] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 83.635281] [<c106c840>] validate_chain+0x320/0x5a0
> [ 83.635281] [<c106df7c>] ? mark_lock+0x21c/0x3c0
> [ 83.635281] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 83.635281] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 83.635281] [<e08150b0>] ? brlock1_lock_init+0xb0/0xb0 [locktest]
> [ 83.635281] [<e08154d0>] ? brlock1_global_unlock+0xa0/0xa0 [locktest]
> [ 83.635281] [<e08150e3>] brlock1_local_lock+0x33/0x90 [locktest]
> [ 83.635281] [<e08150b0>] ? brlock1_lock_init+0xb0/0xb0 [locktest]
> [ 83.635281] [<e08154e8>] locktest_open1+0x18/0x40 [locktest]
> [ 83.635281] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 83.635281] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 83.635281] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 83.635281] [<c11182f0>] ? proc_reg_mmap+0x80/0x80
> [ 83.635281] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 83.635281] [<c10db00c>] ? path_init+0x2cc/0x3c0
> [ 83.635281] [<c10dc846>] path_openat+0xa6/0x340
> [ 83.635281] [<c106d80b>] ? trace_hardirqs_off+0xb/0x10
> [ 83.635281] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 83.635281] [<c13b3a5d>] ? _raw_spin_unlock+0x1d/0x20
> [ 83.635281] [<c10e9f11>] ? alloc_fd+0xe1/0x1a0
> [ 83.635281] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 83.635281] [<c10cfc8b>] ? vfs_write+0x10b/0x130
> [ 83.635281] [<c10cf069>] sys_open+0x29/0x40
> [ 83.635281] [<c13b43c1>] syscall_call+0x7/0xb
>
>
>
> [ 82.758647]
> [ 82.758649] =======================================================
> [ 82.762520] [ INFO: possible circular locking dependency detected ]
> [ 82.762520] 2.6.39-rc3-00228-gd733ed6-dirty #259
> [ 82.762520] -------------------------------------------------------
> [ 82.762520] cat/2768 is trying to acquire lock:
> [ 82.762520] (rwlock1){++++..}, at: [<e081559d>] locktest_open3+0x1d/0x40 [locktest]
> [ 82.762520]
> [ 82.762520] but task is already holding lock:
> [ 82.762520] (&(&(&seqlock1)->lock)->rlock){+.+...}, at: [<e081558d>] locktest_open3+0xd/0x40 [locktest]
> [ 82.762520]
> [ 82.762520] which lock already depends on the new lock.
> [ 82.762520]
> [ 82.762520]
> [ 82.762520] the existing dependency chain (in reverse order) is:
> [ 82.762520]
> [ 82.762520] -> #1 (&(&(&seqlock1)->lock)->rlock){+.+...}:
> [ 82.841627] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 82.841627] [<c106c840>] validate_chain+0x320/0x5a0
> [ 82.841627] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 82.841627] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 82.841627] [<e081560a>] locktest_open4+0x4a/0x90 [locktest]
> [ 82.841627] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 82.841627] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 82.841627] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 82.841627] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 82.841627] [<c10dc846>] path_openat+0xa6/0x340
> [ 82.841627] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 82.841627] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 82.841627] [<c10cf069>] sys_open+0x29/0x40
> [ 82.841627] [<c13b43c1>] syscall_call+0x7/0xb
> [ 82.841627]
> [ 82.841627] -> #0 (rwlock1){++++..}:
> [ 82.841627] [<c106c34c>] check_prev_add+0x78c/0x820
> [ 82.841627] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 82.841627] [<c106c840>] validate_chain+0x320/0x5a0
> [ 82.841627] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 82.841627] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 82.841627] [<c13b3ba9>] _raw_read_lock+0x39/0x70
> [ 82.841627] [<e081559d>] locktest_open3+0x1d/0x40 [locktest]
> [ 82.841627] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 82.841627] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 82.841627] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 82.841627] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 82.841627] [<c10dc846>] path_openat+0xa6/0x340
> [ 82.841627] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 82.841627] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 82.841627] [<c10cf069>] sys_open+0x29/0x40
> [ 82.841627] [<c13b43c1>] syscall_call+0x7/0xb
> [ 82.841627]
> [ 82.841627] other info that might help us debug this:
> [ 82.841627]
> [ 82.841627] 1 lock held by cat/2768:
> [ 82.841627] #0: (&(&(&seqlock1)->lock)->rlock){+.+...}, at: [<e081558d>] locktest_open3+0xd/0x40 [locktest]
> [ 82.841627]
> [ 82.841627] stack backtrace:
> [ 82.841627] Pid: 2768, comm: cat Not tainted 2.6.39-rc3-00228-gd733ed6-dirty #259
> [ 82.841627] Call Trace:
> [ 82.841627] [<c106ade6>] print_circular_bug+0xc6/0xd0
> [ 82.841627] [<c106c34c>] check_prev_add+0x78c/0x820
> [ 82.841627] [<c1005d3b>] ? print_context_stack+0x3b/0xa0
> [ 82.841627] [<c1004fa1>] ? dump_trace+0x81/0xe0
> [ 82.841627] [<c106c499>] check_prevs_add+0xb9/0x110
> [ 82.841627] [<c106c840>] validate_chain+0x320/0x5a0
> [ 82.841627] [<c106df7c>] ? mark_lock+0x21c/0x3c0
> [ 82.841627] [<c106e927>] __lock_acquire+0x2a7/0x8f0
> [ 82.841627] [<c107001a>] lock_acquire+0x7a/0xa0
> [ 82.841627] [<e081559d>] ? locktest_open3+0x1d/0x40 [locktest]
> [ 82.841627] [<e0815580>] ? locktest_open2+0x70/0x70 [locktest]
> [ 82.841627] [<c13b3ba9>] _raw_read_lock+0x39/0x70
> [ 82.841627] [<e081559d>] ? locktest_open3+0x1d/0x40 [locktest]
> [ 82.841627] [<e081559d>] locktest_open3+0x1d/0x40 [locktest]
> [ 82.841627] [<c1118355>] proc_reg_open+0x65/0xe0
> [ 82.841627] [<c10ce78f>] __dentry_open+0x16f/0x2e0
> [ 82.841627] [<c10ce9fe>] nameidata_to_filp+0x5e/0x70
> [ 82.841627] [<c11182f0>] ? proc_reg_mmap+0x80/0x80
> [ 82.841627] [<c10dc1d8>] do_last+0xf8/0x6c0
> [ 82.841627] [<c10db00c>] ? path_init+0x2cc/0x3c0
> [ 82.841627] [<c10dc846>] path_openat+0xa6/0x340
> [ 82.841627] [<c106d80b>] ? trace_hardirqs_off+0xb/0x10
> [ 82.841627] [<c10dcb10>] do_filp_open+0x30/0x80
> [ 82.841627] [<c13b3a5d>] ? _raw_spin_unlock+0x1d/0x20
> [ 82.841627] [<c10e9f11>] ? alloc_fd+0xe1/0x1a0
> [ 82.841627] [<c10cefa1>] do_sys_open+0x101/0x1a0
> [ 82.841627] [<c10cfc8b>] ? vfs_write+0x10b/0x130
> [ 82.841627] [<c10cf069>] sys_open+0x29/0x40
> [ 82.841627] [<c13b43c1>] syscall_call+0x7/0xb
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
--
Only stand for myself
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists