Bjorn, Please rebase your patch on 2.6.38-rc1 and resubmit. We will review it and ACK or NACK the changes, your proposed changes make sense and seem reasonable but we need to apply and test patches before we can commit the changes. Thank you, -Michael Benz -----Original Message----- From: BjÃrn Mork [mailto:bjorn@xxxxxxx] Sent: Tuesday, January 18, 2011 10:34 PM To: Yang, Bo Cc: linux-scsi@xxxxxxxxxxxxxxx; DL-MegaRAID Linux; James E.J. Bottomley Subject: Re: [PATCH] [SCSI] megaraid_sas: Sanity check user supplied length before passing it to dma_alloc_coherent() "Yang, Bo" <Bo.Yang@xxxxxxx> writes: > If you are using megasas ioctl routine to develop your owner > application, the best way you may need to contact with LSI for the > tech support. ??? Thanks for answering, but I'm afraid I don't understand. You don't fix bugs in the driver because you don't support the application used to trigger it? The application in question is smartctl which is part of http://smartmontools.sourceforge.net and AFAIK included with most (all?) Linux distributions. The action required to trigger the driver bug is attempting to initiate a SMART device self test. Quoting from the Debian bug report http://bugs.debian.org/604627 : calling 'smartctl -d megaraid,0 /dev/sda -t short' gives: smartctl 5.40 2010-07-12 r3124 [x86_64-unknown-linux-gnu] (local build) Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net Short offline self test failed [Cannot allocate memory] megasas: Failed to alloc kernel SGL buffer for IOCTL Another user reported the same issue on the linux-ide list: http://marc.info/?l=linux-ide&m=128941801715301 with the following backtrace: [ 69.162393] ------------[ cut here ]------------ [ 69.162404] WARNING: at /build/buildd/linux-2.6.32/mm/page_alloc.c:1806 \ __alloc_pages_slowpath+0x43b/0x580() [ 69.162407] Hardware name: X8DTN [ 69.162409] Modules linked in: fbcon tileblit font bitblit softcursor vga16fb \ vgastate ioatdma radeon ttm drm_kms_helper shpchp drm i2c_algo_bit lp parport floppy \ pata_jmicron megaraid_sas igb dca [ 69.162429] Pid: 1206, comm: smartctl Not \ tainted 2.6.32-25-server #45-Ubuntu [ 69.162432] Call Trace: [ 69.162439] [<ffffffff81065f3b>] warn_slowpath_common+0x7b/0xc0 [ 69.162443] [<ffffffff81065f94>] warn_slowpath_null+0x14/0x20 [ 69.162447] [<ffffffff810f98fb>] __alloc_pages_slowpath+0x43b/0x580 [ 69.162454] [<ffffffff8101078c>] ? __switch_to+0x1ac/0x320 [ 69.162459] [<ffffffff81057850>] ? finish_task_switch+0x50/0xe0 [ 69.162463] [<ffffffff810f9bb1>] __alloc_pages_nodemask+0x171/0x180 [ 69.162468] [<ffffffff81017536>] dma_generic_alloc_coherent+0xa6/0x160 [ 69.162475] [<ffffffff81038b01>] x86_swiotlb_alloc_coherent+0x31/0x70 [ 69.162482] [<ffffffffa002d0ce>] megasas_mgmt_fw_ioctl+0x1ae/0x690 [megaraid_sas] [ 69.162488] [<ffffffffa002d748>] megasas_mgmt_ioctl_fw+0x198/0x240 [megaraid_sas] [ 69.162494] [<ffffffffa002f695>] megasas_mgmt_ioctl+0x35/0x50 [megaraid_sas] [ 69.162500] [<ffffffff81153b12>] vfs_ioctl+0x22/0xa0 [ 69.162505] [<ffffffff8115da2a>] ? alloc_fd+0x10a/0x150 [ 69.162509] [<ffffffff81153cb1>] do_vfs_ioctl+0x81/0x410 [ 69.162515] [<ffffffff8155cc13>] ? do_page_fault+0x153/0x3b0 [ 69.162518] [<ffffffff811540c1>] sys_ioctl+0x81/0xa0 [ 69.162523] [<ffffffff810121b2>] system_call_fastpath+0x16/0x1b [ 69.162526] ---[ end trace 6a2181b634e2abc6 ]--- [ 69.162538] ------------[ cut here ]------------ [ 69.162806] kernel BUG at /build/buildd/linux-2.6.32/lib/swiotlb.c:368! [ 69.163134] invalid opcode: 0000 [#1] SMP [ 69.163570] last sysfs file: \ /sys/devices/system/cpu/cpu3/cache/index2/shared_cpu_map [ 69.163975] CPU 0 [ 69.164227] Modules linked in: fbcon tileblit font bitblit softcursor vga16fb \ vgastate ioatdma radeon ttm drm_kms_helper shpchp drm i2c_algo_bit lp parport floppy \ pata_jmicron megaraid_sas igb dca [ 69.167419] Pid: 1206, comm: smartctl Tainted: G \ W 2.6.32-25-server #45-Ubuntu X8DTN [ 69.167843] RIP: 0010:[<ffffffff812c4dc5>] \ [<ffffffff812c4dc5>] map_single+0x255/0x260 [ 69.168370] RSP: 0018:ffff88081c0ebc58 \ EFLAGS: 00010246 [ 69.168655] RAX: 000000000003bffc RBX: 00000000ffffffff RCX: \ 0000000000000002 [ 69.169000] RDX: 0000000000000000 RSI: 0000000000000000 RDI: \ ffff88001dffe000 [ 69.169346] RBP: ffff88081c0ebcb8 R08: 0000000000000000 R09: \ ffff880000030840 [ 69.169691] R10: 0000000000100000 R11: 0000000000000000 R12: \ 0000000000000000 [ 69.170036] R13: 00000000ffffffff R14: 0000000000000001 R15: \ 0000000000200000 [ 69.170382] FS: 00007fb8de189720(0000) GS:ffff88001de00000(0000) \ knlGS:0000000000000000 [ 69.170794] CS: 0010 DS: 0000 ES: 0000 CR0: \ 0000000080050033 [ 69.171094] CR2: 00007fb8dd59237c CR3: 000000081a790000 CR4: \ 00000000000006f0 [ 69.171439] DR0: 0000000000000000 DR1: 0000000000000000 DR2: \ 0000000000000000 [ 69.171784] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: \ 0000000000000400 [ 69.172130] Process smartctl (pid: 1206, threadinfo \ ffff88081c0ea000, task ffff88081a760000) [ 69.194513] Stack: [ 69.205788] 0000000000000034 00000002817e3390 0000000000000000 ffff88081c0ebe00 [ 69.217739] <0> 0000000000000000 000000000003bffc 0000000000000000 \ 0000000000000000 [ 69.241250] <0> 0000000000000000 00000000ffffffff \ ffff88081c5b4080 ffff88081c0ebe00 [ 69.277310] Call Trace: [ 69.289278] [<ffffffff812c52ac>] swiotlb_alloc_coherent+0xec/0x130 [ 69.301118] [<ffffffff81038b31>] x86_swiotlb_alloc_coherent+0x61/0x70 [ 69.313045] [<ffffffffa002d0ce>] megasas_mgmt_fw_ioctl+0x1ae/0x690 [megaraid_sas] [ 69.336399] [<ffffffffa002d748>] megasas_mgmt_ioctl_fw+0x198/0x240 [megaraid_sas] [ 69.359346] [<ffffffffa002f695>] megasas_mgmt_ioctl+0x35/0x50 [megaraid_sas] [ 69.370902] [<ffffffff81153b12>] vfs_ioctl+0x22/0xa0 [ 69.382322] [<ffffffff8115da2a>] ? alloc_fd+0x10a/0x150 [ 69.393622] [<ffffffff81153cb1>] do_vfs_ioctl+0x81/0x410 [ 69.404696] [<ffffffff8155cc13>] ? do_page_fault+0x153/0x3b0 [ 69.415761] [<ffffffff811540c1>] sys_ioctl+0x81/0xa0 [ 69.426640] [<ffffffff810121b2>] system_call_fastpath+0x16/0x1b [ 69.437491] Code: fe ff ff 48 8b 3d 74 38 76 00 41 bf 00 00 20 00 e8 51 f5 d7 ff \ 83 e0 ff 48 05 ff 07 00 00 48 c1 e8 0b 48 89 45 c8 e9 13 fe ff ff <0f> 0b eb fe 0f 1f \ 80 00 00 00 00 55 48 89 e5 48 83 ec 20 4c 89 [ 69.478216] RIP [<ffffffff812c4dc5>] \ map_single+0x255/0x260 [ 69.489668] RSP <ffff88081c0ebc58> [ 69.500975] ---[ end trace 6a2181b634e2abc7 ]--- Both users have confirmed that the patch fixes their problem. One could of course imagine a workaround in the smartctl application so that it never sent requests with a zero iov_len, but I still believe that actually fixing the driver to handle such requests is better. No? BjÃrn ÿô.nlj·Ÿ®‰†+%ŠË±é¥Šwÿº{.nlj·¥Š{±þÇ‹ø¡Ü}©ž²ÆzÚj:+v‰¨þø®w¥þŠàÞ¨è&¢)ß«a¶Úÿûz¹ÞúŽŠÝjÿŠwèf
