gmane.linux.scsi

[PATCH] libiscsi REGRESSION: fix passthrough support with older iscsi tools

2008-12-01T18:13:00

From: Mike Christie cs.wisc.edu> This regression was added in 2.6.27, when the mtask and ctask were merged into the the common task struct. The patch applies to scsi-rc-fixes, but also applies to 2.6.27 with some offsets. The problem is that __iscsi_conn_send_pdu assumes that userspace was not sending nops with the format it is checking for in the "if" below. It turns out that older userspace tools are. This patch moves the setting of the internal ping_task tracker (it tracks libiscsi current outstanding nop) to iscsi_send_nopout which is only used by kernel callers. Signed-off-by: Mike Christie cs.wisc.edu> --- drivers/scsi/libiscsi.c | 11 +++++------ 1 files changed, 5 insertions(+), 6 deletions(-) diff --git a/drivers/scsi/libiscsi.c b/drivers/scsi/libiscsi.c index 801c7cf..3fdee73 100644 --- a/drivers/scsi/libiscsi.c +++ b/drivers/scsi/libiscsi.c < at >< at > -489,12 +489,6 < at >< at > __iscsi_conn_send_pdu(struct iscsi_conn *conn, struct iscsi_hdr *hdr, if (!__kfifo_get(session->cmdpool.queu

[PATCH] SCSI: fix /proc memory leak in the SCSI core

Alan Stern — 2008-12-01T15:26:48

The SCSI core calls scsi_proc_hostdir_add() from within scsi_host_alloc(), but the corresponding scsi_proc_hostdir_rm() routine is called from within scsi_remove_host(). As a result, if a host is allocated and then deallocated without ever being registered, the host's directory in /proc is leaked. This patch (as1181) fixes this bug in the SCSI core by moving scsi_proc_hostdir_add() into scsi_host_add(). Signed-off-by: Alan Stern rowland.harvard.edu> --- Can somebody check to make sure this is compatible with legacy drivers using the non-hotplug API? Index: usb-2.6/drivers/scsi/hosts.c =================================================================== --- usb-2.6.orig/drivers/scsi/hosts.c +++ usb-2.6/drivers/scsi/hosts.c < at >< at > -206,9 +206,10 < at >< at > int scsi_add_host(struct Scsi_Host *shos if (error) goto fail; +scsi_proc_hostdir_add(shost->hostt); + if (!shost->shost_gendev.parent) shost->shost_gendev.parent = dev ? dev : &platform_bus; - error = device_add(&shost->shost_gendev); if

[PATCH 7/7] remove timeout from struct scsi_device

James Bottomley — 2008-11-30T16:39:01

by removing the unused timeout parameter we ensure a compile failure if anyone is accidentally still using it rather than the block timeout. Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/include/scsi/scsi_device.h b/include/scsi/scsi_device.h index a37a814..2a1c141 100644 --- a/include/scsi/scsi_device.h +++ b/include/scsi/scsi_device.h < at >< at > -159,8 +159,6 < at >< at > struct scsi_device { atomic_t iodone_cnt; atomic_t ioerr_cnt; -int timeout; - struct devicesdev_gendev, sdev_dev; -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 6/7] stex: switch to block timeout

James Bottomley — 2008-11-30T16:38:08

stex sets the timeout in its slave configure routine for all devices. This now needs to update the request queue timeout in block. Cc: Ed Lin promise.com> Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/drivers/scsi/stex.c b/drivers/scsi/stex.c index fd3d4f5..a3a18ad 100644 --- a/drivers/scsi/stex.c +++ b/drivers/scsi/stex.c < at >< at > -477,7 +477,7 < at >< at > stex_slave_config(struct scsi_device *sdev) { sdev->use_10_for_rw = 1; sdev->use_10_for_ms = 1; -sdev->timeout = 60 * HZ; +blk_queue_rq_timeout(sdev->request_queue, 60 * HZ); sdev->tagged_supported = 1; return 0; -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 5/7] st: update to use block timeout

James Bottomley — 2008-11-30T16:36:26

Since we're trying to eliminate struct scsi_device timeout, the tape driver has to be updated to use the block queue timeout instead. The tape use of scsi_device timeout looks to be self consistent, so I don't think this necessarily fixes any bug, but it has to be done to allow me to remove the timeout parameter from struct scsi_device. Cc: Kai Makisara kolumbus.fi> Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/drivers/scsi/st.c b/drivers/scsi/st.c index c959bdc..d60be83 100644 --- a/drivers/scsi/st.c +++ b/drivers/scsi/st.c < at >< at > -613,7 +613,8 < at >< at > static int cross_eof(struct scsi_tape * STp, int forward) tape_name(STp), forward ? "forward" : "backward")); SRpnt = st_do_scsi(NULL, STp, cmd, 0, DMA_NONE, - STp->device->timeout, MAX_RETRIES, 1); + STp->device->request_queue->rq_timeout, + MAX_RETRIES, 1); if (!SRpnt) return (STp->buffer)->syscall_result; < at >< at > -657,7 +658,8 < at >< at > static int st_flush_write_buffer(struct scsi_tape

[PATCH 4/7] make scsi_eh_try_stu use block timeout

James Bottomley — 2008-11-30T16:32:26

scsi_eh_try_stu() was still using the timeout parameter in the device which is now not set (i.e. zero filled) meaning that it waited no time at all for the start unit command to complete (leading the routine to conclude failure every time). This lead to a 2.6.27 regression: http://bugzilla.kernel.org/show_bug.cgi?id=12120 Where firewire devices that were non spec compliant wouldn't spin up. Fix this by using the block queue timeout value instead. Reported-by: Stefan Richter s5r6.in-berlin.de> Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index d07bb3e..381838e 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c < at >< at > -933,8 +933,7 < at >< at > static int scsi_eh_try_stu(struct scsi_cmnd *scmd) int i, rtn = NEEDS_RETRY; for (i = 0; rtn == NEEDS_RETRY && i < 2; i++) -rtn = scsi_send_eh_cmnd(scmd, stu_command, 6, -scmd->device->timeout, 0); +rtn = scsi_send_eh_cmnd(scmd, stu_

[PATCH 3/7] megaraid_sas: switch to block timeout

James Bottomley — 2008-11-30T16:27:51

megaraid_sas sets the timeout in its slave configure routine for devices on special channels. This now needs to update the request queue timeout in block. Cc: "Yang, Bo" lsi.com> Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/drivers/scsi/megaraid/megaraid_sas.c b/drivers/scsi/megaraid/megaraid_sas.c index a454f94..17ce7ab 100644 --- a/drivers/scsi/megaraid/megaraid_sas.c +++ b/drivers/scsi/megaraid/megaraid_sas.c < at >< at > -1016,7 +1016,8 < at >< at > static int megasas_slave_configure(struct scsi_device *sdev) * The RAID firmware may require extended timeouts. */ if (sdev->channel >= MEGASAS_MAX_PD_CHANNELS) -sdev->timeout = MEGASAS_DEFAULT_CMD_TIMEOUT * HZ; +blk_queue_rq_timeout(sdev->request_queue, + MEGASAS_DEFAULT_CMD_TIMEOUT * HZ); return 0; } -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 2/7] ibmvscsi: switch to block timeout

James Bottomley — 2008-11-30T16:20:37

ibmvscsi sets the timeout in its slave configure routine for disk devices. This now needs to update the request queue timeout in block. Cc: Brian King linux.vnet.ibm.com> Signed-off-by: James Bottomley HansenPartnership.com> --- diff --git a/drivers/scsi/ibmvscsi/ibmvscsi.c b/drivers/scsi/ibmvscsi/ibmvscsi.c index 87e09f3..6cad175 100644 --- a/drivers/scsi/ibmvscsi/ibmvscsi.c +++ b/drivers/scsi/ibmvscsi/ibmvscsi.c < at >< at > -1442,7 +1442,7 < at >< at > static int ibmvscsi_slave_configure(struct scsi_device *sdev) spin_lock_irqsave(shost->host_lock, lock_flags); if (sdev->type == TYPE_DISK) { sdev->allow_restart = 1; -sdev->timeout = 60 * HZ; +blk_queue_rq_timeout(sdev->request_queue, 60 * HZ); } scsi_adjust_queue_depth(sdev, 0, shost->cmd_per_lun); spin_unlock_irqrestore(shost->host_lock, lock_flags); -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/major

[PATCH 1/7] aacraid: switch to block timeout

James Bottomley — 2008-11-30T16:15:37

aacraid updates the timeout in its slave configure routine if it is too small. This now needs to update the request queue timeout in block. Signed-off-by: James Bottomley HansenPartnership.com> Cc: AACRAID list adaptec.com> --- diff --git a/drivers/scsi/aacraid/linit.c b/drivers/scsi/aacraid/linit.c index 1c5ac39..9f6f157 100644 --- a/drivers/scsi/aacraid/linit.c +++ b/drivers/scsi/aacraid/linit.c < at >< at > -427,8 +427,8 < at >< at > static int aac_slave_configure(struct scsi_device *sdev) * Firmware has an individual device recovery time typically * of 35 seconds, give us a margin. */ -if (sdev->timeout < (45 * HZ)) -sdev->timeout = 45 * HZ; +if (sdev->request_queue->rq_timeout < (45 * HZ)) +blk_queue_rq_timeout(sdev->request_queue, 45*HZ); for (cid = 0; cid < aac->maximum_num_containers; ++cid) if (aac->fsa_dev[cid].valid) ++num_lsu; -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org

[PATCH 0/7] fix block timeout residue problems

James Bottomley — 2008-11-30T16:12:25

It looks like there was only a partial conversion of the SCSI layer to the block timeout. The missing piece was killing timeout in struct scsi_device and leaving it with a zero value. This has already resulted in a regression: http://bugzilla.kernel.org/show_bug.cgi?id=12120 But on closer inspection, there were lots of other dangling driver uses of the timeout value which would likewise have introduced hard to trace regressions. This patch series eliminates the timeout variable from struct scsi_device and makes everything uniformly use the block timeout. Any wrong use of the scsi device timeout will now result in a compile failure. It's getting late in the -rc series to push this through, but hopefully it won't delay 2.6.28 too much. James -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 00/11] st: remove scsi_execute_async usage (the first half)

FUJITA Tomonori — 2008-11-30T08:07:54

This patchset removes the majority of scsi_execute_async in st driver. IOW, this converts st driver to use the block layer functions. We are in the process of removing scsi_execute_async and scsi_req_map_sg. scsi_execute_async in sg driver were removed in 2.6.27. Now only st and osst drivers use scsi_execute_async(). st driver uses scsi_execute_async for two purposes, performing sort management SCSI commands internally and data transfer between user and kernel space (st_read and st_write). This patchset converts the former. The former performs SCSI commands synchronously. It uses a liner in-kernel buffer (not scatter gather) for data transfer. To perform such commands, other upper layer drivers (e.g. sd) use scsi_execute_req (internally uses blk_rq_map_kern and and blk_execute_rq). scsi_execute_req is not fully fit for st so I adds a helper function similar to scsi_execute_req and replace scsi_execute_async in st driver (I might modify scsi_execute_req for st and remove the helper function later). I'm st

[PATCH] [SCSI] libsas: fix test for negative unsigned and typos

roel kluin — 2008-11-29T16:31:08

unsigned req->data_len cannot be negative, and fix typo Signed-off-by: Roel Kluin gmail.com> --- I am not sure whether this is what was intended, please review. diff --git a/drivers/scsi/libsas/sas_host_smp.c b/drivers/scsi/libsas/sas_host_smp.c index 16f9312..6eb0779 100644 --- a/drivers/scsi/libsas/sas_host_smp.c +++ b/drivers/scsi/libsas/sas_host_smp.c < at >< at > -199,12 +199,12 < at >< at > int sas_smp_host_handler(struct Scsi_Host *shost, struct request *req, break; case SMP_DISCOVER: -req->data_len =- 16; -if (req->data_len < 0) { +if (req->data_len < 15) { req->data_len = 0; error = -EINVAL; goto out; } +req->data_len -= 16; resp_data_len -= 56; sas_host_smp_discover(sas_ha, resp_data, req_data[9]); break; < at >< at > -215,12 +215,12 < at >< at > int sas_smp_host_handler(struct Scsi_Host *shost, struct request *req, break; case SMP_REPORT_PHY_SATA: -req->data_len =- 16; -if (req->data_len < 0) { +if (req->data_len < 15) { req->data_len = 0; error = -EINVAL; got

[Bug 12120] New: [Block layer or SCSI] requests aborted too early during check_partition()

2008-11-29T13:19:55

http://bugzilla.kernel.org/show_bug.cgi?id=12120 Summary: [Block layer or SCSI] requests aborted too early during check_partition() Product: IO/Storage Version: 2.5 KernelVersion: 2.6.28-rc Platform: All OS/Version: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: SCSI AssignedTo: linux-scsi< at >vger.kernel.org ReportedBy: stefan-r-bz< at >s5r6.in-berlin.de OtherBugsDependingO 11808 nThis: Regression: 1 Latest working kernel version: 2.6.27.4 Earliest failing kernel version: 2.6.28-rc6 (I didn't test earlier -rcs yet) Hardware Environment: OxSemi 912 and 92x based FireWire harddisks See http://marc.info/?l=linux-scsi&m=122774022130397 -- Disks may still spin their motor up when the kernel attempts to read partitions. In kernel 2.6.27, this was handled transparently. But since 2.6.28-rc?, the kernel immediately aborts th

[Bug 12119] New: iscsi repeatedly times out

2008-11-29T11:10:39

http://bugzilla.kernel.org/show_bug.cgi?id=12119 Summary: iscsi repeatedly times out Product: IO/Storage Version: 2.5 KernelVersion: 2.6.27.7 Platform: All OS/Version: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: SCSI AssignedTo: linux-scsi< at >vger.kernel.org ReportedBy: ian< at >beware.dropbear.id.au Latest working kernel version:2.6.24.4-64 Earliest failing kernel version:2.6.27.5 Distribution:fc8 Hardware Environment:diskless i386, swap on iscsi device Software Environment:open-iscsi v 6.2 Problem Description:As soon as the iscsi device is activated, the following messages are logged: Nov 29 21:24:51 dhcp-0-20-34-211-135-63 iscsid: Nop-out timedout after 15 seconds on connection 1:0 state (3). Dropping session. Nov 29 21:24:55 dhcp-0-20-34-211-135-63 iscsid: connection1:0 is operational after recovery (2 attempts) These repeat precisely every 30 seconds es

[PATCH] block: integrate blk_end_bidi_request into blk_end_request

FUJITA Tomonori — 2008-11-28T05:52:22

Hi Jens, FC people have been working on FC pass thru feature using bsg bidi support. Seems that the bidi API confuse them: http://marc.info/?l=linux-scsi&m=122704347209856&w=2 = From: FUJITA Tomonori lab.ntt.co.jp> [PATCH] block: integrate blk_end_bidi_request with blk_end_request This integrates blk_end_bidi_request into blk_end_request. IOW, it changes blk_end_request to handle both bidi and non-bidi requests and removes blk_end_bidi_request. Currently, we have two functions to complete a request, blk_end_bidi_request and blk_end_request. The former is for bidi requests and the latter is for non-bidi. This seems to confuse developers. Questions like "can blk_end_bidi_request be used with non-bidi requests?", "what should be passed as the bidi_bytes argument in blk_end_bidi_request" are often asked. The callers don't care about whether a request is bidi or not. All they want to do is to complete a request. I think that a single function that can complete any request would be easy to u

[PATCH 2.6.28-rc6] block: internal dequeue shouldn't start timer

Tejun Heo — 2008-11-28T04:05:16

blkdev_dequeue_request() and elv_dequeue_request() are equivalent and both start the timeout timer. Barrier code dequeues the original barrier request but doesn't passes the request itself to lower level driver, only broken down proxy requests; however, as the original barrier code goes through the same dequeue path and timeout timer is started on it. If barrier sequence takes long enough, this timer expires but the low level driver has no idea about this request and oops follows. Timeout timer shouldn't have been started on the original barrier request as it never goes through actual IO. This patch unexports elv_dequeue_request(), which has no external user anyway, and makes it operate on elevator proper w/o adding the timer and make blkdev_dequeue_request() call elv_dequeue_request() and add timer. Internal users which don't pass the request to driver - barrier code and end_that_request_last() - are converted to use elv_dequeue_request(). Signed-off-by: Tejun Heo kernel.org> Cc: Mike Anderson

[patch 0/1] zfcp fixes for 2.6.28-rc6: missing patch

Christof Schmitt — 2008-11-27T10:44:06

James, this fix should be part of yesterday's patch series for 2.6.28-rc6, but it fell through some holes. It applies cleanly on top of the patch series, please also include this fix. thanks, Christof -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

Bug about mke2fs hang on AIC79 device

Zhang, Yanmin — 2008-11-27T02:18:38

James, 2.6.28-rc1 introduced a bug about scsi lib. We reported it at http://bugzilla.kernel.org/show_bug.cgi?id=11898. Patch at http://bugzilla.kernel.org/show_bug.cgi?id=11898#c36 already fixes the bug. But 2.6.28-rc6 has no the patch. Although there is a performance issue, could you push the patch to upstream firstly? It seems the performance issue isn't related to the patch. Thanks, Yanmin -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo< at >vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

sd or block: regression in 2.6.28-rc WRT manage_start_stop

Stefan Richter — 2008-11-26T22:56:19

Hi, I only recently switched from 2.6.27.y to 2.6.28-rc6 and discovered the following regression with some FireWire HDDs (based on the OXUF 922, OXFW 921, OXUF 924DSB SBP-2 to IDE or SATA bridge chips). The background is that the SBP-2 drivers switch sdev->manage_start_stop unconditionally on for all HDDs since Linux 2.6.27. For example, when I unload firewire-sbp2, the disk's spindle is stopped. When I then reload firewire-sbp2, the SCSI stack does not succeed to restart the HDD under Linux 2.6.28-rc6. It worked under 2.6.27.y. This does not happen with some other disks (tested: HDDs based on OXFW 911, PL-3507, TSB42AA9). The difference between the failing disks and the unaffected disks is explained below. === example of a failing session, 2.6.26-rc8 === # modprobe -r firewire-sbp2 Nov 26 23:25:34 mini sd 2:0:0:0: [sdb] Synchronizing SCSI cache Nov 26 23:25:34 mini sd 2:0:0:0: [sdb] Stopping disk Nov 26 23:25:34 mini firewire_sbp2: released fw1.0, target 2:0:0 # modprobe firewire-sbp2 Nov 26 23

[PATCH 7/7] trivial: move a variable from function to if-scope

Bernd Schubert — 2008-11-26T18:32:53

Move a variable from functions scope to if-scope. Signed-off-by: Bernd Schubert q-leap.de> --- drivers/scsi/scsi_error.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) Index: linux-2.6/drivers/scsi/scsi_error.c =================================================================== --- linux-2.6.orig/drivers/scsi/scsi_error.c +++ linux-2.6/drivers/scsi/scsi_error.c < at >< at > -132,7 +132,6 < at >< at > enum blk_eh_timer_return scsi_times_out( { struct scsi_cmnd *scmd = req->special; enum blk_eh_timer_return (*eh_timed_out)(struct scsi_cmnd *); -enum blk_eh_timer_return rtn = eh_timed_out(scmd); scsi_log_completion(scmd, TIMEOUT_ERROR); < at >< at > -152,6 +151,8 < at >< at > enum blk_eh_timer_return scsi_times_out( eh_timed_out = NULL; if (eh_timed_out) { +enum blk_eh_timer_return rtn = eh_timed_out(scmd); + switch (rtn) { case BLK_EH_NOT_HANDLED: break;

[PATCH 0/7] scsi error handler improvements

Bernd Schubert — 2008-11-26T17:40:44

Hello, over the last year we had an endless number of issues with hard raid units from various vendors. Due this we also have been running into problems with the scsi error handler.