* alignment exceptionhandler sleeps in invalid context @ 2006-04-24 17:32 Olaf Hering 2006-04-24 23:12 ` Paul Mackerras 2006-04-28 11:02 ` Paul Mackerras 0 siblings, 2 replies; 8+ messages in thread From: Olaf Hering @ 2006-04-24 17:32 UTC (permalink / raw) To: linuxppc-dev I'm not sure where the bug is. Does it mean the network stack does something nasty, or is the exception handler itself broken? (probably the latter) This is 2.6.16.9 on a p270. <3>Debug: sleeping function called from invalid context at arch/powerpc/kernel/align.c:440 <4>in_atomic():1, irqs_disabled():0 <4>Call Trace: <4>[C00000000FFE6D10] [C00000000000EE20] .show_stack+0x68/0x1b0 (unreliable) <4>[C00000000FFE6DB0] [C000000000054248] .__might_sleep+0xd8/0xf4 <4>[C00000000FFE6E30] [C00000000000C490] .fix_alignment+0x50c/0x9c8 <4>[C00000000FFE6F10] [C000000000024E20] .alignment_exception+0x18/0xc8 <4>[C00000000FFE6F90] [C00000000000430C] alignment_common+0x10c/0x180 <4>--- Exception: 600 at .csum_partial+0x34/0x9c <4> LR = .skb_checksum+0x194/0x35c <4>[C00000000FFE7280] [D00000000033F3D0] 0xd00000000033f3d0 (unreliable) <4>[C00000000FFE7340] [C0000000002EB558] .skb_checksum_help+0x98/0x124 <4>[C00000000FFE73D0] [D000000000330748] .ip_nat_fn+0x84/0x278 [iptable_nat] <4>[C00000000FFE7490] [D000000000330C34] .ip_nat_local_fn+0x40/0xec [iptable_nat] <4>[C00000000FFE7520] [C00000000030DAE0] .nf_iterate+0x80/0x11c <4>[C00000000FFE75E0] [C00000000030DEC8] .nf_hook_slow+0x8c/0x17c <4>[C00000000FFE76B0] [C00000000031D2B0] .ip_queue_xmit+0x4bc/0x5d0 <4>[C00000000FFE7800] [C00000000032FBA0] .tcp_transmit_skb+0x7d8/0x844 <4>[C00000000FFE78B0] [C0000000003318F0] .__tcp_push_pending_frames+0x2d4/0x3fc <4>[C00000000FFE7970] [C00000000032EAD4] .tcp_rcv_established+0x894/0x964 <4>[C00000000FFE7A20] [C0000000003355C8] .tcp_v4_do_rcv+0x5c/0x48c <4>[C00000000FFE7AF0] [C000000000338A48] .tcp_v4_rcv+0xc78/0xcfc <4>[C00000000FFE7BC0] [C000000000316310] .ip_local_deliver+0x208/0x36c <4>[C00000000FFE7C50] [C000000000316024] .ip_rcv+0x600/0x6e4 <4>[C00000000FFE7CF0] [C0000000002EBB34] .netif_receive_skb+0x500/0x598 <4>[C00000000FFE7D90] [C0000000002EE37C] .process_backlog+0xcc/0x1c8 <4>[C00000000FFE7E40] [C0000000002EE55C] .net_rx_action+0xe4/0x230 <4>[C00000000FFE7EF0] [C000000000066524] .__do_softirq+0x98/0x164 <4>[C00000000FFE7F90] [C000000000025854] .call_do_softirq+0x14/0x24 <4>[C0000003BA3C3370] [C00000000000BCC8] .do_softirq+0x8c/0xd8 <4>[C0000003BA3C3400] [C0000000000663E4] .local_bh_enable+0x58/0x8c <4>[C0000003BA3C3480] [C0000000002EEA34] .dev_queue_xmit+0x2f0/0x318 <4>[C0000003BA3C3510] [C00000000031CA60] .ip_output+0x348/0x3f0 <4>[C0000003BA3C35B0] [C00000000031D2E0] .ip_queue_xmit+0x4ec/0x5d0 <4>[C0000003BA3C3700] [C00000000032FBA0] .tcp_transmit_skb+0x7d8/0x844 <4>[C0000003BA3C37B0] [C0000000003231A8] .tcp_cleanup_rbuf+0x148/0x168 <4>[C0000003BA3C3840] [C000000000325B24] .tcp_recvmsg+0x8a8/0xa00 <4>[C0000003BA3C3930] [C0000000002E0C64] .sock_common_recvmsg+0x5c/0x84 <4>[C0000003BA3C39C0] [C0000000002DD408] .do_sock_read+0x120/0x14c <4>[C0000003BA3C3A60] [C0000000002DF034] .sock_aio_read+0x58/0x74 <4>[C0000003BA3C3B70] [C0000000000C6050] .do_sync_read+0xd4/0x130 <4>[C0000003BA3C3CF0] [C0000000000C6ECC] .vfs_read+0x134/0x1fc <4>[C0000003BA3C3D90] [C0000000000C7390] .sys_read+0x4c/0x8c <4>[C0000003BA3C3E30] [C00000000000871C] syscall_exit+0x0/0x40 ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-24 17:32 alignment exceptionhandler sleeps in invalid context Olaf Hering @ 2006-04-24 23:12 ` Paul Mackerras 2006-04-28 11:02 ` Paul Mackerras 1 sibling, 0 replies; 8+ messages in thread From: Paul Mackerras @ 2006-04-24 23:12 UTC (permalink / raw) To: Olaf Hering; +Cc: linuxppc-dev Olaf Hering writes: > I'm not sure where the bug is. Does it mean the network stack does > something nasty, or is the exception handler itself broken? (probably the latter) > This is 2.6.16.9 on a p270. The alignment handler does a get_user to read the instruction. I suppose it will have to read the instruction directly if the exception occurred in kernel mode. Paul. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-24 17:32 alignment exceptionhandler sleeps in invalid context Olaf Hering 2006-04-24 23:12 ` Paul Mackerras @ 2006-04-28 11:02 ` Paul Mackerras 2006-04-28 11:19 ` Olaf Hering ` (2 more replies) 1 sibling, 3 replies; 8+ messages in thread From: Paul Mackerras @ 2006-04-28 11:02 UTC (permalink / raw) To: Olaf Hering; +Cc: linuxppc-dev Olaf Hering writes: > I'm not sure where the bug is. Does it mean the network stack does > something nasty, or is the exception handler itself broken? (probably the latter) > This is 2.6.16.9 on a p270. This patch should fix it, I hope. If you can verify that it fixes it I'll send it to Linus. Paul. diff --git a/include/asm-powerpc/uaccess.h b/include/asm-powerpc/uaccess.h index 3872e92..b02d858 100644 --- a/include/asm-powerpc/uaccess.h +++ b/include/asm-powerpc/uaccess.h @@ -179,7 +179,8 @@ do { \ #define __put_user_nocheck(x, ptr, size) \ ({ \ long __pu_err; \ - might_sleep(); \ + if ((unsigned long)ptr < PAGE_OFFSET) \ + might_sleep(); \ __chk_user_ptr(ptr); \ __put_user_size((x), (ptr), (size), __pu_err); \ __pu_err; \ @@ -259,7 +260,8 @@ ({ \ long __gu_err; \ unsigned long __gu_val; \ __chk_user_ptr(ptr); \ - might_sleep(); \ + if ((unsigned long)ptr < PAGE_OFFSET) \ + might_sleep(); \ __get_user_size(__gu_val, (ptr), (size), __gu_err); \ (x) = (__typeof__(*(ptr)))__gu_val; \ __gu_err; \ @@ -271,7 +273,8 @@ ({ \ long __gu_err; \ long long __gu_val; \ __chk_user_ptr(ptr); \ - might_sleep(); \ + if ((unsigned long)ptr < PAGE_OFFSET) \ + might_sleep(); \ __get_user_size(__gu_val, (ptr), (size), __gu_err); \ (x) = (__typeof__(*(ptr)))__gu_val; \ __gu_err; \ ^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-28 11:02 ` Paul Mackerras @ 2006-04-28 11:19 ` Olaf Hering 2006-04-28 14:22 ` Kumar Gala 2006-04-29 2:34 ` Michael Ellerman 2006-05-02 4:00 ` Paul Mackerras 2 siblings, 1 reply; 8+ messages in thread From: Olaf Hering @ 2006-04-28 11:19 UTC (permalink / raw) To: Paul Mackerras; +Cc: linuxppc-dev On Fri, Apr 28, Paul Mackeras wrote: > Olaf Hering writes: > > > I'm not sure where the bug is. Does it mean the network stack does > > something nasty, or is the exception handler itself broken? (probably the latter) > > This is 2.6.16.9 on a p270. > > This patch should fix it, I hope. If you can verify that it fixes it > I'll send it to Linus. I dont have a testcase, it just happend once on the p270. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-28 11:19 ` Olaf Hering @ 2006-04-28 14:22 ` Kumar Gala 2006-04-28 23:41 ` Paul Mackerras 0 siblings, 1 reply; 8+ messages in thread From: Kumar Gala @ 2006-04-28 14:22 UTC (permalink / raw) To: Olaf Hering; +Cc: linuxppc-dev, Paul Mackerras On Apr 28, 2006, at 6:19 AM, Olaf Hering wrote: > On Fri, Apr 28, Paul Mackeras wrote: > >> Olaf Hering writes: >> >>> I'm not sure where the bug is. Does it mean the network stack does >>> something nasty, or is the exception handler itself broken? >>> (probably the latter) >>> This is 2.6.16.9 on a p270. >> >> This patch should fix it, I hope. If you can verify that it fixes it >> I'll send it to Linus. > > I dont have a testcase, it just happend once on the p270. I was going to ask how this was actually hit. It seems odd that we would have an alignment error in the network stack. From the oops can we determine what the instruction was that faulted and what address it was trying to access. - k ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-28 14:22 ` Kumar Gala @ 2006-04-28 23:41 ` Paul Mackerras 0 siblings, 0 replies; 8+ messages in thread From: Paul Mackerras @ 2006-04-28 23:41 UTC (permalink / raw) To: Kumar Gala; +Cc: linuxppc-dev, Olaf Hering Kumar Gala writes: > I was going to ask how this was actually hit. It seems odd that we > would have an alignment error in the network stack. From the oops > can we determine what the instruction was that faulted and what > address it was trying to access. No, it's not particularly odd, given that our get_unaligned() and put_unaligned() in include/asm-powerpc/unaligned.h just do a single access. That makes the common case (address is actually aligned, or cpu handles it in hardware) fast at the expense of occasionally taking an alignment exception. Paul. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-28 11:02 ` Paul Mackerras 2006-04-28 11:19 ` Olaf Hering @ 2006-04-29 2:34 ` Michael Ellerman 2006-05-02 4:00 ` Paul Mackerras 2 siblings, 0 replies; 8+ messages in thread From: Michael Ellerman @ 2006-04-29 2:34 UTC (permalink / raw) To: Paul Mackerras; +Cc: linuxppc-dev, Olaf Hering [-- Attachment #1: Type: text/plain, Size: 1144 bytes --] On Fri, 2006-04-28 at 21:02 +1000, Paul Mackerras wrote: > Olaf Hering writes: > > > I'm not sure where the bug is. Does it mean the network stack does > > something nasty, or is the exception handler itself broken? (probably the latter) > > This is 2.6.16.9 on a p270. > > This patch should fix it, I hope. If you can verify that it fixes it > I'll send it to Linus. > > Paul. > > diff --git a/include/asm-powerpc/uaccess.h b/include/asm-powerpc/uaccess.h > index 3872e92..b02d858 100644 > --- a/include/asm-powerpc/uaccess.h > +++ b/include/asm-powerpc/uaccess.h > @@ -179,7 +179,8 @@ do { \ > #define __put_user_nocheck(x, ptr, size) \ > ({ \ > long __pu_err; \ > - might_sleep(); \ > + if ((unsigned long)ptr < PAGE_OFFSET) \ > + might_sleep(); \ + if (!is_kernel_addr((unsigned long)ptr)) \ + might_sleep(); \ In asm/page.h :) cheers -- Michael Ellerman IBM OzLabs wwweb: http://michael.ellerman.id.au phone: +61 2 6212 1183 (tie line 70 21183) We do not inherit the earth from our ancestors, we borrow it from our children. - S.M.A.R.T Person [-- Attachment #2: This is a digitally signed message part --] [-- Type: application/pgp-signature, Size: 191 bytes --] ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: alignment exceptionhandler sleeps in invalid context 2006-04-28 11:02 ` Paul Mackerras 2006-04-28 11:19 ` Olaf Hering 2006-04-29 2:34 ` Michael Ellerman @ 2006-05-02 4:00 ` Paul Mackerras 2 siblings, 0 replies; 8+ messages in thread From: Paul Mackerras @ 2006-05-02 4:00 UTC (permalink / raw) To: Olaf Hering, linuxppc-dev I wrote: > This patch should fix it, I hope. If you can verify that it fixes it > I'll send it to Linus. but init barfs at startup with that patch, due to the double evaluation of `ptr'. This one runs a bit better... Paul. diff --git a/include/asm-powerpc/uaccess.h b/include/asm-powerpc/uaccess.h index 3872e92..d83fc29 100644 --- a/include/asm-powerpc/uaccess.h +++ b/include/asm-powerpc/uaccess.h @@ -7,6 +7,7 @@ #ifndef __ASSEMBLY__ #include <linux/sched.h> #include <linux/errno.h> #include <asm/processor.h> +#include <asm/page.h> #define VERIFY_READ 0 #define VERIFY_WRITE 1 @@ -179,9 +180,11 @@ do { \ #define __put_user_nocheck(x, ptr, size) \ ({ \ long __pu_err; \ - might_sleep(); \ - __chk_user_ptr(ptr); \ - __put_user_size((x), (ptr), (size), __pu_err); \ + __typeof__(*(ptr)) __user *__pu_addr = (ptr); \ + if (!is_kernel_addr((unsigned long)__pu_addr)) \ + might_sleep(); \ + __chk_user_ptr(ptr); \ + __put_user_size((x), __pu_addr, (size), __pu_err); \ __pu_err; \ }) @@ -258,9 +261,11 @@ #define __get_user_nocheck(x, ptr, size) ({ \ long __gu_err; \ unsigned long __gu_val; \ + const __typeof__(*(ptr)) __user *__gu_addr = (ptr); \ __chk_user_ptr(ptr); \ - might_sleep(); \ - __get_user_size(__gu_val, (ptr), (size), __gu_err); \ + if (!is_kernel_addr((unsigned long)__gu_addr)) \ + might_sleep(); \ + __get_user_size(__gu_val, __gu_addr, (size), __gu_err); \ (x) = (__typeof__(*(ptr)))__gu_val; \ __gu_err; \ }) @@ -270,9 +275,11 @@ #define __get_user64_nocheck(x, ptr, siz ({ \ long __gu_err; \ long long __gu_val; \ - __chk_user_ptr(ptr); \ - might_sleep(); \ - __get_user_size(__gu_val, (ptr), (size), __gu_err); \ + const __typeof__(*(ptr)) __user *__gu_addr = (ptr); \ + __chk_user_ptr(ptr); \ + if (!is_kernel_addr((unsigned long)__gu_addr)) \ + might_sleep(); \ + __get_user_size(__gu_val, __gu_addr, (size), __gu_err); \ (x) = (__typeof__(*(ptr)))__gu_val; \ __gu_err; \ }) ^ permalink raw reply related [flat|nested] 8+ messages in thread
end of thread, other threads:[~2006-05-02 4:00 UTC | newest] Thread overview: 8+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2006-04-24 17:32 alignment exceptionhandler sleeps in invalid context Olaf Hering 2006-04-24 23:12 ` Paul Mackerras 2006-04-28 11:02 ` Paul Mackerras 2006-04-28 11:19 ` Olaf Hering 2006-04-28 14:22 ` Kumar Gala 2006-04-28 23:41 ` Paul Mackerras 2006-04-29 2:34 ` Michael Ellerman 2006-05-02 4:00 ` Paul Mackerras
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).