| 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
| ||
|
Date: Wed, 2 Jun 2021 10:43:29 -0700
From: Kees Cook <keescook@...omium.org>
To: "Zhengyejian (Zetta)" <zhengyejian1@...wei.com>,
Joel Fernandes <joel@...lfernandes.org>
Cc: rostedt@...dmis.org, ccross@...roid.com,
linux-kernel@...r.kernel.org, Zhangjinhao <zhangjinhao2@...wei.com>
Subject: Re: [BUG] I found a bug when try to enable record_ftrace
On Wed, Jun 02, 2021 at 03:42:23PM +0800, Zhengyejian (Zetta) wrote:
> Hello,
>
> There may be a deadlock caused by ftrace recursion when try to enable
> record_ftrace.
Hi,
Thanks for the report. Joel, is this something you can take a look at?
-Kees
> I'd like to known if the patchset
> (https://lore.kernel.org/lkml/20201106023235.367190737@goodmis.org/) is
> trying to fix it.
>
> Procedure for reproducing the problem is:
> 1) this problem appears both in 5.13-rc4 and 5.10;
>
> 2) my work environment is:
> qemu-arm version 4.0.0
> arm-linux-gnueabi-gcc (Ubuntu/Linaro 7.5.0-3ubuntu1~18.04) 7.5.0
>
> 3) then try to enable record_ftrace:
> / # mount -t pstore pstore /sys/fs/pstore
> / # echo 1 > /sys/kernel/debug/pstore/record_ftrace
>
> 4) then system appears to be stuck, and use arm-linux-gnueabi-gdb dump the
> following call stack:
> #0 arch_spin_lock (lock=0x811a0f98) at
> /home/zyj/Linux/linux-master/arch/arm/include/asm/spinlock.h:74
> #1 do_raw_spin_lock_flags (flags=<synthetic pointer>, lock=0x811a0f98)
> at /home/zyj/Linux/linux-master/include/linux/spinlock.h:195
> #2 __raw_spin_lock_irqsave (lock=0x811a0f98) at
> /home/zyj/Linux/linux-master/include/linux/spinlock_api_smp.h:119
> #3 _raw_spin_lock_irqsave (lock=lock@...ry=0x811a0f98) at
> /home/zyj/Linux/linux-master/kernel/locking/spinlock.c:159
> #4 0x8046c6e0 in buffer_size_add (prz=prz@...ry=0x811a0f80,
> a=a@...ry=16) at /home/zyj/Linux/linux-master/fs/pstore/ram_core.c:82
> #5 0x8046cc20 in persistent_ram_write (prz=0x811a0f80, s=0x81137b80,
> count=16) at /home/zyj/Linux/linux-master/fs/pstore/ram_core.c:327
> #6 0x8046b438 in ramoops_pstore_write (record=0x81137b90) at
> /home/zyj/Linux/linux-master/fs/pstore/ram.c:331
> #7 0x8046add8 in pstore_ftrace_call (ip=2156609456,
> parent_ip=2152122068, op=<optimized out>, fregs=<optimized out>)
> at /home/zyj/Linux/linux-master/fs/pstore/ftrace.c:54
> #8 0x801dd580 in __ftrace_ops_list_func (ignored=0x0, fregs=0x0,
> parent_ip=2152122068, ip=2156609456) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:7003
> #9 ftrace_ops_list_func (ip=2156609456, parent_ip=2152122068,
> op=<optimized out>, fregs=0x0) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:7028
> #10 0x801109f8 in ftrace_caller () at
> /home/zyj/Linux/linux-master/arch/arm/kernel/entry-ftrace.S:224
> #11 0x808b3fb4 in _raw_spin_unlock_irqrestore
> (lock=lock@...ry=0x811a0f98, flags=1610612883) at
> /home/zyj/Linux/linux-master/kernel/locking/spinlock.c:190
> #12 0x8046c6d4 in buffer_size_add (prz=prz@...ry=0x811a0f80,
> a=a@...ry=16) at /home/zyj/Linux/linux-master/fs/pstore/ram_core.c:95
> #13 0x8046cc20 in persistent_ram_write (prz=0x811a0f80, s=0x81137cf8,
> count=16) at /home/zyj/Linux/linux-master/fs/pstore/ram_core.c:327
> #14 0x8046b438 in ramoops_pstore_write (record=0x81137d08) at
> /home/zyj/Linux/linux-master/fs/pstore/ram.c:331
> #15 0x8046add8 in pstore_ftrace_call (ip=2148632188,
> parent_ip=2148601660, op=<optimized out>, fregs=<optimized out>)
> at /home/zyj/Linux/linux-master/fs/pstore/ftrace.c:54
> #16 0x801dd580 in __ftrace_ops_list_func (ignored=0x0, fregs=0x0,
> parent_ip=2148601660, ip=2148632188) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:7003
> #17 ftrace_ops_list_func (ip=2148632188, parent_ip=2148601660,
> op=<optimized out>, fregs=0x0) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:7028
> #18 0x801109f8 in ftrace_caller () at
> /home/zyj/Linux/linux-master/arch/arm/kernel/entry-ftrace.S:224
> #19 0x80118680 in __set_fixmap (idx=idx@...ry=FIX_TEXT_POKE0,
> phys=phys@...ry=0, prot=prot@...ry=0) at
> /home/zyj/Linux/linux-master/arch/arm/mm/mmu.c:385
> #20 0x80110f3c in patch_unmap (flags=<synthetic pointer>, fixmap=129) at
> /home/zyj/Linux/linux-master/arch/arm/kernel/patch.c:45
> #21 __patch_text_real (addr=addr@...ry=0x808b3fb0
> <_raw_spin_unlock_irqrestore+16>, insn=insn@...ry=3957420680,
> remap=remap@...ry=true)
> at /home/zyj/Linux/linux-master/arch/arm/kernel/patch.c:104
> #22 0x80110b40 in __patch_text (insn=3957420680, addr=0x808b3fb0
> <_raw_spin_unlock_irqrestore+16>) at
> /home/zyj/Linux/linux-master/arch/arm/include/asm/patch.h:10
> #23 ftrace_modify_code (pc=2156609456, old=3904716800, new=3957420680,
> validate=true) at /home/zyj/Linux/linux-master/arch/arm/kernel/ftrace.c:95
> #24 0x80110cf4 in ftrace_make_call (rec=rec@...ry=0x81007ce0,
> addr=addr@...ry=2148600280) at
> /home/zyj/Linux/linux-master/arch/arm/kernel/ftrace.c:132
> #25 0x801dda40 in __ftrace_replace_code (rec=rec@...ry=0x81007ce0,
> enable=enable@...ry=true) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:2524
> #26 0x801de17c in ftrace_replace_code (mod_flags=mod_flags@...ry=1) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:2554
> #27 0x801de39c in ftrace_modify_all_code (command=5) at
> /home/zyj/Linux/linux-master/kernel/trace/ftrace.c:2698
> #28 0x80110ad8 in __ftrace_modify_code (data=<optimized out>) at
> /home/zyj/Linux/linux-master/arch/arm/kernel/ftrace.c:39
> #29 0x801d4da4 in multi_cpu_stop (data=data@...ry=0x81c69d84) at
> /home/zyj/Linux/linux-master/kernel/stop_machine.c:240
> #30 0x801d4a14 in cpu_stopper_thread (cpu=<optimized out>) at
> /home/zyj/Linux/linux-master/kernel/stop_machine.c:511
> #31 0x8014de3c in smpboot_thread_fn (data=0x810c1c80) at
> /home/zyj/Linux/linux-master/kernel/smpboot.c:165
> #32 0x8014a0dc in kthread (_create=0x810c1d40) at
> /home/zyj/Linux/linux-master/kernel/kthread.c:313
> #33 0x80100150 in ret_from_fork () at
> /home/zyj/Linux/linux-master/arch/arm/kernel/entry-common.S:158
> Backtrace stopped: previous frame identical to this frame (corrupt
> stack?)
>
> See above #7~#15, there is a recursion in function pstore_ftrace_call(), and
> a spin_lock(lock=0x811a0f98) is hold since unlock operation in above #11 not
> finished. Then in above #0, trying to acquire same lock cause a deadlock.
>
> Enabling 'record_ftrace' seems a basic operation of pstore/ftrace, Does it
> mean this feature is not available for a while?
>
> Best Regards,
> Zheng
--
Kees Cook
Powered by blists - more mailing lists