-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
This is with 2.6.0-test1-ac3, on a Sony PCG-R505TFP.
This has an i830 chipset, and
02:02.0 FireWire (IEEE 1394): Texas Instruments TSB43AB22 1394a-2000 Controller
Compiler: gcc version 3.2 (Mandrake Linux 9.0 3.2-1mdk)
Version_in_Makefile: 2.6.0-test1-ac3
CONFIG_IEEE1394=y
# CONFIG_IEEE1394_VERBOSEDEBUG is not set
# CONFIG_IEEE1394_OUI_DB is not set
CONFIG_IEEE1394_OHCI1394=y
# CONFIG_IEEE1394_VIDEO1394 is not set
CONFIG_IEEE1394_SBP2=y
# CONFIG_IEEE1394_SBP2_PHYS_DMA is not set
# CONFIG_IEEE1394_ETH1394 is not set
# CONFIG_IEEE1394_DV1394 is not set
# CONFIG_IEEE1394_RAWIO is not set
# CONFIG_IEEE1394_CMP is not set
CONFIG_DEBUG_KERNEL=y
CONFIG_DEBUG_STACKOVERFLOW=y
# CONFIG_DEBUG_SLAB is not set
CONFIG_DEBUG_IOVIRT=y
CONFIG_MAGIC_SYSRQ=y
CONFIG_DEBUG_SPINLOCK=y
# CONFIG_DEBUG_PAGEALLOC is not set
CONFIG_DEBUG_SPINLOCK_SLEEP=y
# CONFIG_FRAME_POINTER is not set
CONFIG_X86_EXTRA_IRQS=y
CONFIG_X86_FIND_SMP_CONFIG=y
CONFIG_X86_MPPARSE=y
Debug: sleeping function called from invalid context at include/asm/semaphore.h:119
Call Trace:
[<c011c61e>] __might_sleep+0x5e/0x62
[<c031bfad>] hpsb_get_tlabel+0x5d/0x230
[<c0319d97>] alloc_hpsb_packet+0xa7/0xd0
[<c031c4e2>] hpsb_make_writepacket+0xa2/0x140
[<c01283e6>] update_wall_time+0x16/0x40
[<c032a270>] sbp2util_allocate_write_packet+0x40/0x80
[<c032c5c6>] sbp2_link_orb_command+0x86/0x190
[<c032c773>] sbp2_send_command+0xa3/0xf0
[<c032cd70>] sbp2scsi_queuecommand+0xb0/0x210
[<c02ff550>] scsi_done+0x0/0x80
[<c0301b30>] scsi_times_out+0x0/0x60
[<c02ff33b>] scsi_dispatch_cmd+0x18b/0x250
[<c02ff550>] scsi_done+0x0/0x80
[<c0301b30>] scsi_times_out+0x0/0x60
[<c03037fd>] scsi_init_cmd_errh+0x9d/0xd0
[<c0304ad1>] scsi_request_fn+0x201/0x460
[<c02b5133>] __elv_add_request+0x33/0x50
[<c02b78d4>] blk_insert_request+0x84/0x100
[<c0303599>] scsi_do_req+0x49/0xa0
[<c03034a3>] scsi_insert_special_req+0x33/0x40
[<c0303731>] scsi_wait_req+0x71/0xa0
[<c03035f0>] scsi_wait_done+0x0/0xd0
[<c0300322>] ioctl_internal_command+0x52/0x190
[<c03004c6>] scsi_set_medium_removal+0x66/0x90
[<c0147001>] invalidate_inode_pages+0x21/0x30
[<c032e11b>] cdrom_release+0x6b/0x130
[<c0312a80>] sr_block_release+0x0/0x20
[<c0166029>] blkdev_put+0x1f9/0x240
[<c01662ba>] close_bdev_excl+0x1a/0x30
[<c016443c>] kill_block_super+0x3c/0x50
[<c0162ebd>] deactivate_super+0x8d/0x140
[<c017cd65>] __mntput+0x25/0x40
[<c017d74c>] sys_umount+0x3c/0xa0
[<c017d7c9>] sys_oldumount+0x19/0x20
[<c01099d3>] syscall_call+0x7/0xb
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
iD8DBQE/InLsW6pHgIdAuOMRAn+BAJ0bUhhON0riFKcCrcMjrOPFi8MeeQCcCEaN
RWU40vdfHOdjM1BAdspRvj8=
=cW5Q
-----END PGP SIGNATURE-----
I think sbp2scsi_queuecommand is called from outside interrupt context. The
obvious but possibly wrong way to fix this would be to change the calls to
hpsb_get_tlabel() to check in_atomic() instead of in_interrupt().
regards,
dan carpenter
On Saturday 26 July 2003 05:24 am, Brad Hards wrote:
> Debug: sleeping function called from invalid context at
> include/asm/semaphore.h:119 Call Trace:
> [<c011c61e>] __might_sleep+0x5e/0x62
> [<c031bfad>] hpsb_get_tlabel+0x5d/0x230
calls down()
> [<c0319d97>] alloc_hpsb_packet+0xa7/0xd0
calls hpsb_get_tlabel(packet, in_interrupt() ? 0 : 1)
> [<c031c4e2>] hpsb_make_writepacket+0xa2/0x140
calls alloc_hpsb_packet(length + (length % 4 ? 4 - (length % 4) : 0));
calls hpsb_get_tlabel(packet, in_interrupt() ? 0 : 1)
> [<c032c5c6>] sbp2_link_orb_command+0x86/0x190
calls hpsb_make_writepacket()
> [<c032c773>] sbp2_send_command+0xa3/0xf0
calls sbp2util_allocate_command_orb(scsi_id, SCpnt, done);
> [<c032cd70>] sbp2scsi_queuecommand+0xb0/0x210
calls sbp2_send_command under a spinlock
On Sun, Jul 27, 2003 at 03:46:50AM -0700, dan carpenter wrote:
> I think sbp2scsi_queuecommand is called from outside interrupt context. The
> obvious but possibly wrong way to fix this would be to change the calls to
> hpsb_get_tlabel() to check in_atomic() instead of in_interrupt().
Preempt will be the death of me. Wish there was a can_sleep() macro to
make things easier.
Thanks for the report, I'll fix this.
--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
On Sun, Jul 27, 2003 at 03:46:50AM -0700, dan carpenter wrote:
> I think sbp2scsi_queuecommand is called from outside interrupt context. The
> obvious but possibly wrong way to fix this would be to change the calls to
> hpsb_get_tlabel() to check in_atomic() instead of in_interrupt().
This is fixed in the SVN repo now. I'll push to Linus soon.
--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/