lkml.org 
[lkml]   [2014]   [Sep]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRCU bug with v3.17-rc3 ?
Hi,

I keep triggering the following Oops with -rc3 when writing to the mass
storage gadget driver:

| # modprobe g_mass_storage stall=0 removable=1 file=/dev/sda
| [ 44.883554] Number of LUNs=8
| [ 44.886709] Mass Storage Function, version: 2009/09/11
| [ 44.892303] LUN: removable file: (no medium)
| [ 44.896916] Number of LUNs=1
| [ 44.901198] LUN: removable file: /dev/sda
| [ 44.905410] Number of LUNs=1
| [ 44.917706] g_mass_storage gadget: Mass Storage Gadget, version: 2009/09/11
| [ 44.925018] g_mass_storage gadget: userspace failed to provide iSerialNumber
| [ 44.932489] g_mass_storage gadget: g_mass_storage ready
| [ 52.583773] g_mass_storage gadget: high-speed config #1: Linux File-Backed Storage
| # [ 98.270585] Unable to handle kernel paging request at virtual address ffffffff
| [ 98.278198] pgd = c0004000
| [ 98.281027] [ffffffff] *pgd=ae7f6821, *pte=00000000, *ppte=00000000
| [ 98.287648] Internal error: Oops: 17 [#1] SMP ARM
| [ 98.292559] Modules linked in: g_mass_storage usb_f_mass_storage libcomposite configfs usb_storage xhci_hcd dwc3 udc_core matrix_keypad lis3lv02d_i2c dwc3_omap lis3lv02d input_polldev
| [ 98.309721] CPU: 0 PID: 1820 Comm: file-storage Not tainted 3.17.0-rc3-00013-gc6b1a7d #806
| [ 98.318346] task: ec356040 ti: ec378000 task.ti: ec378000
| [ 98.324000] PC is at find_get_entry+0x7c/0x128
| [ 98.328640] LR is at 0xfffffffa
| [ 98.331912] pc : [<c011394c>] lr : [<fffffffa>] psr: a0000013
| [ 98.331912] sp : ec379b50 ip : 00000000 fp : ec379b84
| [ 98.343888] r10: c0c81243 r9 : 00000001 r8 : ea123d28
| [ 98.349352] r7 : ec378010 r6 : 00000001 r5 : 00000000 r4 : 0000000f
| [ 98.356181] r3 : ec379b3c r2 : 00000000 r1 : 00000001 r0 : ffffffff
| [ 98.363006] Flags: NzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment kernel
| [ 98.370646] Control: 10c5387d Table: ac2b0059 DAC: 00000015
| [ 98.376641] Process file-storage (pid: 1820, stack limit = 0xec378248)
| [ 98.383454] Stack: (0xec379b50 to 0xec37a000)
| [ 98.388003] 9b40: 00000000 00000000 c01138d0 c002aa3c
| [ 98.396560] 9b60: 0000000f 00000000 ea123d24 000200d0 00000001 000000d0 ec379bbc ec379b88
| [ 98.405100] 9b80: c0114360 c01138dc c1486a00 60000013 ec379bc4 00001400 00000000 ea123d24
| [ 98.413635] 9ba0: 00000c00 00000400 ec378010 c06dea0c ec379bdc ec379bc0 c011478c c0114330
| [ 98.422183] 9bc0: 000000d0 c00904f8 c1486a00 00001400 ec379c04 ec379be0 c019cd68 c0114760
| [ 98.430732] 9be0: c0090808 c0090590 ec379c34 00000001 00000c00 ea123d24 ec379c2c ec379c08
| [ 98.439300] 9c00: c019ecbc c019cd44 00000c00 00000001 ec379c58 c019eb9c 00000c00 ec379d54
| [ 98.447860] 9c20: ec379c8c ec379c30 c0113f14 c019ec8c 00000c00 00000001 ec379c58 ec379c5c
| [ 98.456414] 9c40: ec378030 00000001 ec250cc0 00000000 00001400 00000000 c018195c c00acd08
| [ 98.464974] 9c60: 5408b05a 00001000 ec250cc0 00000000 ec379d68 ea123d24 ec378010 00000000
| [ 98.473533] 9c80: ec379cf4 ec379c90 c0115ed4 c0113e6c 00000001 00000000 c019f2b0 c0090590
| [ 98.482071] 9ca0: ec379cc4 ec378010 c06c3df4 00001000 ea123c64 c019f2b0 ec379d54 ec379cc8
| [ 98.490607] 9cc0: 00001400 00000000 00000001 ec379d68 ec379d54 ec379e30 ec250cc0 ec356040
| [ 98.499178] 9ce0: ed7ab800 ec30d800 ec379d3c ec379cf8 c019f2b0 c0115c8c c06be3b8 c006dcec
| [ 98.507741] 9d00: ec1b0010 ec30d800 ec379d08 ec379d08 ec379d10 ec379d10 ec379d18 ec379d18
| [ 98.516288] 9d20: 00001400 00000000 ec379e30 ec250cc0 ec379dc4 ec379d40 c016618c c019f284
| [ 98.524833] 9d40: 00001000 c0317b78 ec379d7c ec394000 00001000 00000003 00000000 00001000
| [ 98.533385] 9d60: ec379d4c 00000001 ec250cc0 00000000 00000000 00000000 ec356040 00000000
| [ 98.541946] 9d80: 00000000 00000000 00001400 00000000 00001000 00000000 00000000 00000000
| [ 98.550482] 9da0: ec394000 ec250cc0 ec394000 ec379e30 00001000 00001000 ec379df4 ec379dc8
| [ 98.559023] 9dc0: c0166a3c c01660f4 00000002 ec0ace20 00001000 0000000e ec0ace00 00000000
| [ 98.567567] 9de0: 00001000 ed7ab800 ec379e64 ec379df8 bf0bc3b4 c0166994 0000006f 00001000
| [ 98.576112] 9e00: bf0bc7a4 60000013 e8156000 0000000e 3930343d 00000000 bf0bc7a4 ec0ace00
| [ 98.584660] 9e20: 00002400 00000000 00001400 00000000 00001400 00000000 ec379e64 00000000
| [ 98.593193] 9e40: ed36ddc0 ec378018 ec30d894 ec0ace00 ec30d800 ec30d840 ec379ed4 ec379e68
| [ 98.601754] 9e60: bf0bd1c8 bf0bc08c bf0bf6ec ec378010 c06c3df4 ec356040 00000001 00000000
| [ 98.610305] 9e80: ec379eac ec379e90 c00906b0 c00904f8 ec30d894 ed36ddc0 ec378018 ec30d894
| [ 98.618857] 9ea0: ec379ebc ec379eb0 c0090808 ec30d800 ed36ddc0 ec378018 ec30d894 00000000
| [ 98.627405] 9ec0: 00000200 ec0ace00 ec379f14 ec379ed8 bf0bdbe8 bf0bc74c c06c3d94 ec0acc80
| [ 98.635942] 9ee0: ec394000 ec30d800 bf0bd8cc ec0acc80 00000000 ec30d800 bf0bd8cc 00000000
| [ 98.644465] 9f00: 00000000 00000000 ec379fac ec379f18 c0066ac4 bf0bd8d8 ed1d1040 00000000
| [ 98.652990] 9f20: ec379f3c ec30d800 00000000 00000000 dead4ead ffffffff ffffffff c0c86138
| [ 98.661526] 9f40: 00000000 00000000 c08998e0 00000000 c006dd7c ec379f54 ec379f54 00000000
| [ 98.670077] 9f60: 00000000 dead4ead ffffffff ffffffff c0c86138 00000000 00000000 c08998e0
| [ 98.678612] 9f80: 00000000 ec379f90 ec379f88 ec379f88 ec0acc80 c00669e0 00000000 00000000
| [ 98.687148] 9fa0: 00000000 ec379fb0 c000eea8 c00669ec 00000000 00000000 00000000 00000000
| [ 98.695699] 9fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
| [ 98.704249] 9fe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000
| [ 98.712805] [<c011394c>] (find_get_entry) from [<c0114360>] (pagecache_get_page+0x3c/0x1f0)
| [ 98.721529] [<c0114360>] (pagecache_get_page) from [<c011478c>] (grab_cache_page_write_begin+0x38/0x50)
| [ 98.731345] [<c011478c>] (grab_cache_page_write_begin) from [<c019cd68>] (block_write_begin+0x30/0x90)
| [ 98.741067] [<c019cd68>] (block_write_begin) from [<c019ecbc>] (blkdev_write_begin+0x3c/0x48)
| [ 98.749974] [<c019ecbc>] (blkdev_write_begin) from [<c0113f14>] (generic_perform_write+0xb4/0x1e4)
| [ 98.759335] [<c0113f14>] (generic_perform_write) from [<c0115ed4>] (__generic_file_write_iter+0x254/0x51c)
| [ 98.769424] [<c0115ed4>] (__generic_file_write_iter) from [<c019f2b0>] (blkdev_write_iter+0x38/0xc0)
| [ 98.778978] [<c019f2b0>] (blkdev_write_iter) from [<c016618c>] (new_sync_write+0xa4/0xcc)
| [ 98.787526] [<c016618c>] (new_sync_write) from [<c0166a3c>] (vfs_write+0xb4/0x1c0)
| [ 98.795462] [<c0166a3c>] (vfs_write) from [<bf0bc3b4>] (do_write+0x334/0x53c [usb_f_mass_storage])
| [ 98.804858] [<bf0bc3b4>] (do_write [usb_f_mass_storage]) from [<bf0bd1c8>] (do_scsi_command+0xa88/0x118c [usb_f_mass_storage])
| [ 98.816782] [<bf0bd1c8>] (do_scsi_command [usb_f_mass_storage]) from [<bf0bdbe8>] (fsg_main_thread+0x31c/0x72c [usb_f_mass_storage])
| [ 98.829249] [<bf0bdbe8>] (fsg_main_thread [usb_f_mass_storage]) from [<c0066ac4>] (kthread+0xe4/0x100)
| [ 98.838993] [<c0066ac4>] (kthread) from [<c000eea8>] (ret_from_fork+0x14/0x20)
| [ 98.846554] Code: e1a01009 eb0905d4 e3500000 0a00001f (e5904000)
| [ 98.853110] ---[ end trace 8bdf31522b942652 ]---


The setup is a bit "odd", I have a USB stick attached to the host port
on my platform and the peripheral port uses that stick as backing file.
that is connected to a laptop which I'm using to read/write to that
backing file. The problem doesn't seem to trigger if I run the exact
same test straight to the USB stick which is attached to the host port.

My test application is rather basic [1] which I run with a script [2] to
pass sensible arguments. I haven't found another way to reproducing this
yet, so it could very well be that g_mass_storage is at fault here, as I
also managed to trigger this when using a tmpfs as backing file.

Anyway, looking at PC:

| (gdb) list *(find_get_entry+0x7c)
| 0xc011394c is in find_get_entry (include/linux/radix-tree.h:196).
| 191 * radix_tree_deref_retry must be used to confirm validity of the pointer if
| 192 * only the read lock is held.
| 193 */
| 194 static inline void *radix_tree_deref_slot(void **pslot)
| 195 {
| 196 return rcu_dereference(*pslot);
| 197 }
| 198
| 199 /**
| 200 * radix_tree_deref_slot_protected - dereference a slot without RCU lock but with tree lock held
| (gdb)

And looking at the arguments for that function, we're passing r0 as
0xffffffff and r1 as 1, which clearly is bogus, but I don't know, at
least not yet, where did those come from. I'll see if I can reproduce
the same problem with dummy_hcd to rule out a bug in my dwc3 driver :-)

cheers

--
balbi
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2014-09-04 21:01    [W:1.623 / U:0.132 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site