Problem with exception handling - program counter is corrupted
From: Tal Mussayoff <hidden>
Date: 2013-12-10 09:43:25
Hi,
Is there any update? Do you need more information?
Here is a brief description of my problem:
We've recently encountered a problem in the kernel, where the program counter is corrupted, when handling SIGSEGV exception, that has been caused on the user space. We've checked this under kernel version 3.10 (3.12 is not stable).
I've simulated an exception on the code using the following piece of code:
UINT32* pD = NULL;
memset(pD, 0, 10);
With the help of traces I've got the following log:
2013-10-01T18:09:53.508501+00:00 zodiac kernel: Pid: 890, comm: InputPortManage
2013-10-01T18:09:53.520828+00:00 zodiac kernel: CPU: 0 Tainted: G O (3.9.0-0.1.0.0.16xilinx-zynq-g3f7c2d5 #94)
2013-10-01T18:09:53.528178+00:00 zodiac kernel: PC is at 0x76a2f790
2013-10-01T18:09:53.535436+00:00 zodiac kernel: LR is at 0x3d23ad
2013-10-01T18:09:53.549813+00:00 zodiac kernel: pc : [<76a2f790>] lr : [<003d23ad>] psr: 20000010
2013-10-01T18:09:53.549836+00:00 zodiac kernel: sp : 767feb58 ip :
00000000 fp : 00000000
2013-10-01T18:09:53.559259+00:00 zodiac kernel: r10: 76ff5000 r9 :
00000000 r8 : 00000001
2013-10-01T18:09:53.570065+00:00 zodiac kernel: r7 : 767feb60 r6 :
00000001 r5 : 00000001 r4 : 00451e04
2013-10-01T18:09:53.580802+00:00 zodiac kernel: r3 : 00000000 r2 :
00002708 r1 : 00000000 r0 : 00000000
2013-10-01T18:09:53.592253+00:00 zodiac kernel: Flags: nzCv IRQs on
FIQs on Mode USER_32 ISA ARM Segment user
2013-10-01T18:09:53.602230+00:00 zodiac kernel: Control: 18c5387d
Table: 0cd6804a DAC: 00000015
2013-10-01T18:09:53.610594+00:00 zodiac kernel: ~~~~~~ DUMP BACKTRACE
1 ~~~~~~
2013-10-01T18:09:53.619764+00:00 zodiac kernel: ~~~~~~~~~ UNWIND BACKTRACE 4 ~~~~~~~~~~~
2013-10-01T18:09:53.629694+00:00 zodiac kernel: unwind_backtrace(regs = (null) tsk = (null))
2013-10-01T18:09:53.636168+00:00 zodiac kernel: ~~~TSK~~~
2013-10-01T18:09:53.647228+00:00 zodiac kernel: ~~~ frame.fp 896c9dd4,
sp 896c9da8, lr 80018168, pc 80013c94
2013-10-01T18:09:53.660499+00:00 zodiac kernel: [<80013c94>]
(unwind_backtrace+0x0/0x154) from [<80018168>]
(__do_user_fault+0x60/0xdc)
2013-10-01T18:09:53.673678+00:00 zodiac kernel: [<80018168>]
(__do_user_fault+0x60/0xdc) from [<800184ac>]
(do_page_fault+0x2c8/0x350)
2013-10-01T18:09:53.686627+00:00 zodiac kernel: [<800184ac>]
(do_page_fault+0x2c8/0x350) from [<800083f8>] (do_DataAbort+0x34/0x98)
2013-10-01T18:09:53+00:00 zodiac IRP: ConfDeamon: error [Transport
endpoint is not connected] sending message total_size=167
sent_bytes=-1
2013-10-01T18:09:53.699161+00:00 zodiac kernel: [<800083f8>] (do_DataAbort+0x34/0x98) from [<8000e034>] (__dabt_usr+0x34/0x40)
2013-10-01T18:09:53.708400+00:00 zodiac kernel: Exception stack(0x896c9fb0 to 0x896c9ff8)
2013-10-01T18:09:53.720865+00:00 zodiac kernel: 9fa0: 00000000 00000000 00002708 00000000
2013-10-01T18:09:53.733238+00:00 zodiac kernel: 9fc0: 00451e04
00000001 00000001 767feb60 00000001 00000000 76ff5000 00000000
2013-10-01T18:09:53.744067+00:00 zodiac kernel: 9fe0: 00000000
767feb58 003d23ad 76a2f790 20000010 ffffffff
The resulted PC, is invalid one.
For some unknown reason, when I've modified my code to:
*pD = 328;
The problem disappears, and the following logs are printed:
2013-10-01T18:11:38.209091+00:00 zodiac kernel: Pid: 810, comm: InputPortManage 2013-10-01T18:11:38.221395+00:00 zodiac kernel: CPU: 0 Tainted: G O (3.9.0-0.1.0.0.16xilinx-zynq-g3f7c2d5 #92) 2013-10-01T18:11:38.228583+00:00 zodiac kernel: PC is at 0x3d23a2 2013-10-01T18:11:38.235841+00:00 zodiac kernel: LR is at 0x3d2395 2013-10-01T18:11:38.250218+00:00 zodiac kernel: pc : [<003d23a2>] lr : [<003d2395>] psr: 00000030 2013-10-01T18:11:38.250241+00:00 zodiac kernel: sp : 7676ab60 ip : 76a0602c fp : 00000000 2013-10-01T18:11:38.259918+00:00 zodiac kernel: r10: 76f45000 r9 : 00000000 r8 : 00000001 2013-10-01T18:11:38.270471+00:00 zodiac kernel: r7 : 7676ab60 r6 : 00000001 r5 : 00000001 r4 : 00451dfc 2013-10-01T18:11:38.281208+00:00 zodiac kernel: r3 : 00000000 r2 : 00000148 r1 : 00000001 r0 : 00000000 2013-10-01T18:11:38.292866+00:00 zodiac kernel: Flags: nzcv IRQs on FIQs on Mode USER_32 ISA Thumb Segment user 2013-10-01T18:11:38.302820+00:00 zodiac kernel: Control: 18c5387d Table: 034d804a DAC: 00000015 2013-10-01T18:11:38.320353+00:00 zodiac kernel: ~~~~~~ DUMP BACKTRACE 1 ~~~~~~ 2013-10-01T18:11:38.320376+00:00 zodiac kernel: ~~~~~~~~~ UNWIND BACKTRACE 4 ~~~~~~~~~~~ 2013-10-01T18:11:38.330284+00:00 zodiac kernel: unwind_backtrace(regs = (null) tsk = (null)) 2013-10-01T18:11:38.336758+00:00 zodiac kernel: ~~~TSK~~~ 2013-10-01T18:11:38.347818+00:00 zodiac kernel: ~~~ frame.fp 86f53ddc, sp 86f53db0, lr 80018158, pc 80013c94 2013-10-01T18:11:38.361066+00:00 zodiac kernel: [<80013c94>] (unwind_backtrace+0x0/0x154) from [<80018158>] (__do_user_fault+0x50/0xbc) 2013-10-01T18:11:38+00:00 zodiac IRP: DecrMngr-ENGINE: Dispatch: Illegal Msg opcode 0x7 received 2013-10-01T18:11:38.374337+00:00 zodiac kernel: [<80018158>] (__do_user_fault+0x50/0xbc) from [<8001848c>] (do_page_fault+0x2c8/0x350) 2013-10-01T18:11:38.387194+00:00 zodiac kernel: [<8001848c>] (do_page_fault+0x2c8/0x350) from [<800083f8>] (do_DataAbort+0x34/0x98) 2013-10-01T18:11:38.399705+00:00 zodiac kernel: [<800083f8>] (do_DataAbort+0x34/0x98) from [<8000e034>] (__dabt_usr+0x34/0x40) 2013-10-01T18:11:38.408967+00:00 zodiac kernel: Exception stack(0x86f53fb0 to 0x86f53ff8) 2013-10-01T18:11:38.421409+00:00 zodiac kernel: 3fa0: 00000000 00000001 00000148 00000000 2013-10-01T18:11:38.433804+00:00 zodiac kernel: 3fc0: 00451dfc 00000001 00000001 7676ab60 00000001 00000000 76f45000 00000000 2013-10-01T18:11:38.444656+00:00 zodiac kernel: 3fe0: 76a0602c 7676ab60 003d2395 003d23a2 00000030 ffffffff In this case the PC value is valid.
Thanks, Tal. -----Original Message----- From: Tal Mussayoff Sent: Wednesday, December 04, 2013 11:50 To: 'linux-arm-kernel at lists.infradead.org' Cc: 'Michal Simek' Subject: Kernel Problem Hi, We've recently encountered a problem in the kernel, where the program counter is corrupted, when handling SIGSEGV exception, that has been caused on the user space. We've checked this under kernel version 3.10 (3.12 is not stable). Please find more information below. We'll appreciate any help you can supply, Regards, Tal. -----Original Message----- From: Michal Simek [mailto:michal.simek at xilinx.com] Sent: Thursday, November 28, 2013 16:10 To: Tal Mussayoff Cc: michal.simek at xilinx.com; git at Xilinx.com Subject: Re: Kernel Problem Hi, please use git at xilinx.com for questions like this. But anyway this behaviour is not xilinx specific and should be discussed on arm mailing list. Also I would recommend you to try this on 3.12 and if you see the same problem send it to the arm mailing list. Thanks, Michal On 11/28/2013 02:52 PM, Tal Mussayoff wrote:
Hi Michal,
I'm Tal, and I'm working on a project which includes kernel version 3.9, on Harmonic Inc. We've encountered a problem with the exceptions handling, which led to a failure of receiving stack backtrace. I've simulated an exception on the code (SIGSEGV signal) using the following piece of code:
UINT32* pD = NULL;
memset(pD, 0, 10);
). With the help of traces I've got the following log:
2013-10-01T18:09:53.508501+00:00 zodiac kernel: Pid: 890, comm: InputPortManage
2013-10-01T18:09:53.520828+00:00 zodiac kernel: CPU: 0 Tainted: G O (3.9.0-0.1.0.0.16xilinx-zynq-g3f7c2d5 #94)
2013-10-01T18:09:53.528178+00:00 zodiac kernel: PC is at 0x76a2f790
2013-10-01T18:09:53.535436+00:00 zodiac kernel: LR is at 0x3d23ad
2013-10-01T18:09:53.549813+00:00 zodiac kernel: pc : [<76a2f790>] lr : [<003d23ad>] psr: 20000010
2013-10-01T18:09:53.549836+00:00 zodiac kernel: sp : 767feb58 ip :
00000000 fp : 00000000
2013-10-01T18:09:53.559259+00:00 zodiac kernel: r10: 76ff5000 r9 :
00000000 r8 : 00000001
2013-10-01T18:09:53.570065+00:00 zodiac kernel: r7 : 767feb60 r6 :
00000001 r5 : 00000001 r4 : 00451e04
2013-10-01T18:09:53.580802+00:00 zodiac kernel: r3 : 00000000 r2 :
00002708 r1 : 00000000 r0 : 00000000
2013-10-01T18:09:53.592253+00:00 zodiac kernel: Flags: nzCv IRQs on
FIQs on Mode USER_32 ISA ARM Segment user
2013-10-01T18:09:53.602230+00:00 zodiac kernel: Control: 18c5387d
Table: 0cd6804a DAC: 00000015
2013-10-01T18:09:53.610594+00:00 zodiac kernel: ~~~~~~ DUMP BACKTRACE
1 ~~~~~~
2013-10-01T18:09:53.619764+00:00 zodiac kernel: ~~~~~~~~~ UNWIND BACKTRACE 4 ~~~~~~~~~~~
2013-10-01T18:09:53.629694+00:00 zodiac kernel: unwind_backtrace(regs = (null) tsk = (null))
2013-10-01T18:09:53.636168+00:00 zodiac kernel: ~~~TSK~~~
2013-10-01T18:09:53.647228+00:00 zodiac kernel: ~~~ frame.fp 896c9dd4,
sp 896c9da8, lr 80018168, pc 80013c94
2013-10-01T18:09:53.660499+00:00 zodiac kernel: [<80013c94>]
(unwind_backtrace+0x0/0x154) from [<80018168>]
(__do_user_fault+0x60/0xdc)
2013-10-01T18:09:53.673678+00:00 zodiac kernel: [<80018168>]
(__do_user_fault+0x60/0xdc) from [<800184ac>]
(do_page_fault+0x2c8/0x350)
2013-10-01T18:09:53.686627+00:00 zodiac kernel: [<800184ac>]
(do_page_fault+0x2c8/0x350) from [<800083f8>] (do_DataAbort+0x34/0x98)
2013-10-01T18:09:53+00:00 zodiac IRP: ConfDeamon: error [Transport
endpoint is not connected] sending message total_size=167
sent_bytes=-1
2013-10-01T18:09:53.699161+00:00 zodiac kernel: [<800083f8>] (do_DataAbort+0x34/0x98) from [<8000e034>] (__dabt_usr+0x34/0x40)
2013-10-01T18:09:53.708400+00:00 zodiac kernel: Exception stack(0x896c9fb0 to 0x896c9ff8)
2013-10-01T18:09:53.720865+00:00 zodiac kernel: 9fa0: 00000000 00000000 00002708 00000000
2013-10-01T18:09:53.733238+00:00 zodiac kernel: 9fc0: 00451e04
00000001 00000001 767feb60 00000001 00000000 76ff5000 00000000
2013-10-01T18:09:53.744067+00:00 zodiac kernel: 9fe0: 00000000
767feb58 003d23ad 76a2f790 20000010 ffffffff
The resulted PC, is invalid one.
For some unknown reason, when I've modified my code to:
*pD = 328;
The problem disappears, and the following logs are printed:
2013-10-01T18:11:38.209091+00:00 zodiac kernel: Pid: 810, comm: InputPortManage
2013-10-01T18:11:38.221395+00:00 zodiac kernel: CPU: 0 Tainted: G O (3.9.0-0.1.0.0.16xilinx-zynq-g3f7c2d5 #92)
2013-10-01T18:11:38.228583+00:00 zodiac kernel: PC is at 0x3d23a2
2013-10-01T18:11:38.235841+00:00 zodiac kernel: LR is at 0x3d2395
2013-10-01T18:11:38.250218+00:00 zodiac kernel: pc : [<003d23a2>] lr : [<003d2395>] psr: 00000030
2013-10-01T18:11:38.250241+00:00 zodiac kernel: sp : 7676ab60 ip :
76a0602c fp : 00000000
2013-10-01T18:11:38.259918+00:00 zodiac kernel: r10: 76f45000 r9 :
00000000 r8 : 00000001
2013-10-01T18:11:38.270471+00:00 zodiac kernel: r7 : 7676ab60 r6 :
00000001 r5 : 00000001 r4 : 00451dfc
2013-10-01T18:11:38.281208+00:00 zodiac kernel: r3 : 00000000 r2 :
00000148 r1 : 00000001 r0 : 00000000
2013-10-01T18:11:38.292866+00:00 zodiac kernel: Flags: nzcv IRQs on
FIQs on Mode USER_32 ISA Thumb Segment user
2013-10-01T18:11:38.302820+00:00 zodiac kernel: Control: 18c5387d
Table: 034d804a DAC: 00000015
2013-10-01T18:11:38.320353+00:00 zodiac kernel: ~~~~~~ DUMP BACKTRACE
1 ~~~~~~
2013-10-01T18:11:38.320376+00:00 zodiac kernel: ~~~~~~~~~ UNWIND BACKTRACE 4 ~~~~~~~~~~~
2013-10-01T18:11:38.330284+00:00 zodiac kernel: unwind_backtrace(regs = (null) tsk = (null))
2013-10-01T18:11:38.336758+00:00 zodiac kernel: ~~~TSK~~~
2013-10-01T18:11:38.347818+00:00 zodiac kernel: ~~~ frame.fp 86f53ddc,
sp 86f53db0, lr 80018158, pc 80013c94
2013-10-01T18:11:38.361066+00:00 zodiac kernel: [<80013c94>]
(unwind_backtrace+0x0/0x154) from [<80018158>]
(__do_user_fault+0x50/0xbc)
2013-10-01T18:11:38+00:00 zodiac IRP: DecrMngr-ENGINE: Dispatch:
Illegal Msg opcode 0x7 received
2013-10-01T18:11:38.374337+00:00 zodiac kernel: [<80018158>]
(__do_user_fault+0x50/0xbc) from [<8001848c>]
(do_page_fault+0x2c8/0x350)
2013-10-01T18:11:38.387194+00:00 zodiac kernel: [<8001848c>]
(do_page_fault+0x2c8/0x350) from [<800083f8>] (do_DataAbort+0x34/0x98)
2013-10-01T18:11:38.399705+00:00 zodiac kernel: [<800083f8>]
(do_DataAbort+0x34/0x98) from [<8000e034>] (__dabt_usr+0x34/0x40)
2013-10-01T18:11:38.408967+00:00 zodiac kernel: Exception stack(0x86f53fb0 to 0x86f53ff8)
2013-10-01T18:11:38.421409+00:00 zodiac kernel: 3fa0: 00000000 00000001 00000148 00000000
2013-10-01T18:11:38.433804+00:00 zodiac kernel: 3fc0: 00451dfc
00000001 00000001 7676ab60 00000001 00000000 76f45000 00000000
2013-10-01T18:11:38.444656+00:00 zodiac kernel: 3fe0: 76a0602c
7676ab60 003d2395 003d23a2 00000030 ffffffff In this case the PC value is valid.
The problem is pretty much consistent when using memset.
I'm not sure if this is relevant, but the PSR has consistent value for each scenario (0x20000010 when the problem occurs, and 0x00000030 when it doesn't).
We'll appreciate any help,
Thanks,
Tal Mussayoff.
P.S.
As you may have understand, I'm the one who posted a question on the Xilinx forum :).