[PATCH] BNX2I: Fixed kernel panic due to illegal usage of sc-request-cpu
A kernel panic was observed when passing the sc-request-cpu = -1 to retrieve the per_cpu variable pointer: #0 [880011203960] machine_kexec at 81022bc3 #1 [8800112039b0] crash_kexec at 81088630 #2 [880011203a80] __die at 8139ea20 #3 [880011203aa0] no_context at 8102f3a7 #4 [880011203ae0] __bad_area_nosemaphore at 8102f665 #5 [880011203ba0] retint_signal at 8139dd1f #6 [880011203cc8] bnx2i_indicate_kcqe at a03dc4f2 #7 [880011203da8] service_kcqes at a03cb04f #8 [880011203e68] cnic_service_bnx2x_kcq at a03cb14a #9 [880011203e88] cnic_service_bnx2x_bh at a03cb1b3 The problem lies in the sg_io (and perhaps sg_scsi_ioctl) call to blk_get_request-get_request/wait-blk_alloc_request-blk_rq_init which re-initializes the request-cpu to -1. There is no assignment for cpu from that to the request_fn call to low level drivers. When this happens, the sc-request-cpu will be using the init value of -1. This will create a kernel panic when it hits bnx2i because the code refers it to get the per_cpu variables ptr. This change is to put in a guard against that and also for cases when CONFIG_SMP/BIO_CPU_AFFINE is not enabled. In those cases, the cpu affinitization code would not get run in __make_request either; hence the request-cpu will remain a -1 also. This bug was created from commit: b5cf6b63f73abdc051035f0050b367beeb2ef94c Signed-off-by: Eddie Wai eddie@broadcom.com --- drivers/scsi/bnx2i/bnx2i_iscsi.c |4 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/drivers/scsi/bnx2i/bnx2i_iscsi.c b/drivers/scsi/bnx2i/bnx2i_iscsi.c index 5c55a75..622383d 100644 --- a/drivers/scsi/bnx2i/bnx2i_iscsi.c +++ b/drivers/scsi/bnx2i/bnx2i_iscsi.c @@ -1225,6 +1225,10 @@ static int bnx2i_task_xmit(struct iscsi_task *task) if (!sc) return bnx2i_mtask_xmit(conn, task); + if (!blk_rq_cpu_valid(sc-request)) { + sc-request-cpu = get_cpu(); + put_cpu(); + } bnx2i_setup_cmd_wqe_template(cmd); cmd-req.op_code = ISCSI_OP_SCSI_CMD; cmd-conn = bnx2i_conn; -- 1.7.0.5 -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: [PATCH] BNX2I: Fixed kernel panic due to illegal usage of sc-request-cpu
On 07/11/2011 01:14 PM, Eddie Wai wrote: A kernel panic was observed when passing the sc-request-cpu = -1 to retrieve the per_cpu variable pointer: #0 [880011203960] machine_kexec at 81022bc3 #1 [8800112039b0] crash_kexec at 81088630 #2 [880011203a80] __die at 8139ea20 #3 [880011203aa0] no_context at 8102f3a7 #4 [880011203ae0] __bad_area_nosemaphore at 8102f665 #5 [880011203ba0] retint_signal at 8139dd1f #6 [880011203cc8] bnx2i_indicate_kcqe at a03dc4f2 #7 [880011203da8] service_kcqes at a03cb04f #8 [880011203e68] cnic_service_bnx2x_kcq at a03cb14a #9 [880011203e88] cnic_service_bnx2x_bh at a03cb1b3 The problem lies in the sg_io (and perhaps sg_scsi_ioctl) call to blk_get_request-get_request/wait-blk_alloc_request-blk_rq_init which re-initializes the request-cpu to -1. There is no assignment for cpu from that to the request_fn call to low level drivers. When this happens, the sc-request-cpu will be using the init value of -1. This will create a kernel panic when it hits bnx2i because the code refers it to get the per_cpu variables ptr. This change is to put in a guard against that and also for cases when CONFIG_SMP/BIO_CPU_AFFINE is not enabled. In those cases, the cpu affinitization code would not get run in __make_request either; hence the request-cpu will remain a -1 also. diff --git a/drivers/scsi/bnx2i/bnx2i_iscsi.c b/drivers/scsi/bnx2i/bnx2i_iscsi.c index 5c55a75..622383d 100644 --- a/drivers/scsi/bnx2i/bnx2i_iscsi.c +++ b/drivers/scsi/bnx2i/bnx2i_iscsi.c @@ -1225,6 +1225,10 @@ static int bnx2i_task_xmit(struct iscsi_task *task) if (!sc) return bnx2i_mtask_xmit(conn, task); + if (!blk_rq_cpu_valid(sc-request)) { + sc-request-cpu = get_cpu(); + put_cpu(); + } If I understand you right, then I think this needs to get fixed in the block or scsi layer instead of each LLD. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: Implement iSCSI dev_loss_tmo
On 07/09/2011 03:00 PM, Vivek S wrote: Hi, In the TODO list newly added to the open-iscsi repo, kernel item 2 says that the dev_loss_tmo handler needs to be made common. Does this mean that we need to move this into the scsi layer by adding it to scsi_transport_template structure instead of iscsi_cls_session structure (patch submitted at thread http://groups.google.com/group/open-iscsi/msg/031510ab4cecccfd?dmode=source indicates so) ? Exactly. So we want to add a callback function similar to how their is a callback function for the scanning (scsi_host_template-user_scan). We probably then want to think of a way to have the transport class or some common code (maybe in scsi_lib.c) add a sysfs file (dev_loss_tmo) to the transport object that the new callback is managing. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: iSCSI TODO
On 07/11/2011 04:06 AM, Tomoaki Nishimura wrote: # I resend the email, because I didn't receive my email from open-iscsi mailing list. Hi I will take USERSPACE TODO: 3 iscsiadm cannot update a setting of a running session. Regards Ok. Thanks. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: iSCSI TODO
On 07/09/2011 11:42 PM, Aastha Mehta wrote: Hi Mike, Thanks for the reply. I will get started on TODO 1 after my trip. Ok. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.