scsi: core: regression fixes for request batching
Subsystems (5)
cover scsi: core: regression fixes for request batching
Hi James, Martin, Paolo, Ming,
multipathing with linux-next is broken since 20190723 in our CI. The patches fix a memleak and a severe dh/multipath functional regression. It would be nice if we could get them to 5.4/scsi-queue and also next.
I would have preferred if such a new feature had used its own new copy scsi_mq_ops_batching instead of changing the use case and semantics of the existing scsi_mq_ops, because this would likely cause less regressions for all the other users not using the new feature.
Steffen Maier (2):
scsi: core: fix missing .cleanup_rq for SCSI hosts without request
batching
scsi: core: fix dh and multipathing for SCSI hosts without request
batching
drivers/scsi/scsi_lib.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Signature · 1 line
2.17.1
PATCH 1/2 scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
This was missing from scsi_mq_ops_no_commit of linux-next commit 8930a6c20791 ("scsi: core: add support for request batching") from Martin's scsi/5.4/scsi-queue or James' scsi/misc.
See also linux-next commit b7e9e1fb7a92 ("scsi: implement .cleanup_rq callback") from block/for-next.
--- drivers/scsi/scsi_lib.c | 1 + 1 file changed, 1 insertion(+)
drivers/scsi/scsi_lib.c +1 -0
@@ -1834,6 +1834,7 @@ static const struct blk_mq_ops scsi_mq_ops_no_commit = { .init_request = scsi_mq_init_request, .exit_request = scsi_mq_exit_request, .initialize_rq_fn = scsi_initialize_rq, + .cleanup_rq = scsi_cleanup_rq, .busy = scsi_mq_lld_busy, .map_queues = scsi_map_queues, };
Signature · 1 line
2.17.1
Re: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
Quoted 21 lines · Steffen Maier
This was missing from
scsi_mq_ops_no_commitof linux-next commit8930a6c20791("scsi: core: add support for request batching") from Martin's scsi/5.4/scsi-queue or James' scsi/misc.See also linux-next commit
b7e9e1fb7a92("scsi: implement .cleanup_rq callback") fromblock/for-next.Signed-off-by Steffen Maier <[email protected]>Fixes 8930a6c20791 ("scsi: core: add support for request batching")Cc Paolo Bonzini <[email protected]>Cc Ming Lei <[email protected]>--- drivers/scsi/scsi_lib.c | 1 + 1 file changed, 1 insertion(+)drivers/scsi/scsi_lib.c +0
-0
drivers/scsi/scsi_lib.c static const struct blk_mq_ops scsi_mq_ops_no_commit = { · line 1840
Quoted 7 lines
drivers/scsi/scsi_lib.c +1
-0@@ -1834,6 +1834,7 @@ static const struct blk_mq_ops scsi_mq_ops_no_commit = { .init_request = scsi_mq_init_request, .exit_request = scsi_mq_exit_request, .initialize_rq_fn = scsi_initialize_rq, + .cleanup_rq = scsi_cleanup_rq, .busy = scsi_mq_lld_busy, .map_queues = scsi_map_queues, };
This one is a cross-tree thing, either scsi/5.4/scsi-queue needs to pull for-5.4/block, or
do it after both land linus tree.
Thanks,
Ming Lei
Re: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
Ming,
Quoted 7 lines
Quoted 4 lines
+ .cleanup_rq = scsi_cleanup_rq, .busy = scsi_mq_lld_busy, .map_queues = scsi_map_queues,};
This one is a cross-tree thing, either scsi/5.4/scsi-queue needs to pull for-5.4/block, or do it after both land linus tree.
I'll set up an amalgamated for-next branch tomorrow.
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Re: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
Quoted 8 lines · Martin K. Petersen
Ming,
Quoted 5 lines
Quoted 4 lines
+ .cleanup_rq =
scsi_cleanup_rq,
.busy =scsi_mq_lld_busy,
.map_queues =scsi_map_queues,
};
Quoted 4 lines
Quoted 2 lines
This one is a cross-tree thing, either scsi/5.4/scsi-queue needs to pull for-5.4/block, or do it after both land linus tree.
I'll set up an amalgamated for-next branch tomorrow.
Martin, is it possible that you re-wrote your for-next and it now no longer contains a merged 5.4/scsi-postmerge with those fixes? At least I cannot find the fix code in next-20190917 and it fails again for me.
Signature · 11 lines
Mit freundlichen Gruessen / Kind regards Steffen Maier Linux on IBM Z Development https://www.ibm.com/privacy/us/en/ IBM Deutschland Research & Development GmbH Vorsitzender des Aufsichtsrats: Matthias Hartmann Geschaeftsfuehrung: Dirk Wittkopp Sitz der Gesellschaft: Boeblingen Registergericht: Amtsgericht Stuttgart, HRB 243294
Re: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
Steffen,
Quoted 4 lines
Martin, is it possible that you re-wrote your for-next and it now no longer contains a merged 5.4/scsi-postmerge with those fixes? At least I cannot find the fix code in next-20190917 and it fails again for me.
Yes, looks like you're right. Not sure how I managed to mess that up. I must have inadvertently done a reset in the wrong worktree because my for-next branch maintenance script only does merges.
In any case, since Linus has pulled the block tree dependencies, I'll rebase the postmerge branch on top of current linus/master and create a new for-next.
Thanks for the heads-up!
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Mark Brown
maintainerRe: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
In reply to Steffen MaierQuoted 7 lines · Steffen Maier
Quoted 1 line · Martin K. Petersen
I'll set up an amalgamated for-next branch tomorrow.
Martin, is it possible that you re-wrote your for-next and it now no longer contains a merged 5.4/scsi-postmerge with those fixes?
At least I cannot find the fix code in next-20190917 and it fails again for me.
Well, there's no sign of a branch called postmerge in the SCSI history recently and I've not run into any SCSI-related conflicts so...
kbuild test robot
build botRe: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
In reply to Steffen MaierI love your patch! Yet something to improve:
Hi Steffen,
I love your patch! Yet something to improve:
[auto build test ERROR on linus/master]
[cannot apply to v5.3-rc3 next-20190807]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Steffen-Maier/scsi-core-fix-missing-cleanup_rq-for-SCSI-hosts-without-request-batching/20190808-052017
config: riscv-defconfig (attached as .config)
compiler: riscv64-linux-gcc (GCC) 7.4.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.4.0 make.cross ARCH=riscv
If you fix the issue, kindly add following tag
All errors (new ones prefixed by >>):
>> drivers/scsi/scsi_lib.c:1824:3: error: 'const struct blk_mq_ops' has no member named 'cleanup_rq'; did you mean 'queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~
queue_rq
>> drivers/scsi/scsi_lib.c:1824:16: error: 'scsi_cleanup_rq' undeclared here (not in a function); did you mean 'scsi_queue_rq'? .cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~~~~~~ scsi_queue_rq
vim +1824 drivers/scsi/scsi_lib.c
1811
1812 static const struct blk_mq_ops scsi_mq_ops = {
1813 .get_budget = scsi_mq_get_budget,
1814 .put_budget = scsi_mq_put_budget,
1815 .queue_rq = scsi_queue_rq,
1816 .complete = scsi_softirq_done,
1817 .timeout = scsi_timeout,
1818 #ifdef CONFIG_BLK_DEBUG_FS
1819 .show_rq = scsi_show_rq,
1820 #endif
1821 .init_request = scsi_mq_init_request,
1822 .exit_request = scsi_mq_exit_request,
1823 .initialize_rq_fn = scsi_initialize_rq,
> 1824 .cleanup_rq = scsi_cleanup_rq,
1825 .busy = scsi_mq_lld_busy,
1826 .map_queues = scsi_map_queues,
1827 };
1828
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
kbuild test robot
build botRe: [PATCH 1/2] scsi: core: fix missing .cleanup_rq for SCSI hosts without request batching
In reply to Steffen MaierI love your patch! Yet something to improve:
Hi Steffen,
I love your patch! Yet something to improve:
[auto build test ERROR on linus/master]
[cannot apply to v5.3-rc3 next-20190808]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Steffen-Maier/scsi-core-fix-missing-cleanup_rq-for-SCSI-hosts-without-request-batching/20190808-052017
config: s390-allmodconfig (attached as .config)
compiler: s390-linux-gcc (GCC) 7.4.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.4.0 make.cross ARCH=s390
If you fix the issue, kindly add following tag
All errors (new ones prefixed by >>):
drivers//scsi/scsi_lib.c:1824:3: error: 'const struct blk_mq_ops' has no member named 'cleanup_rq'; did you mean 'queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~
queue_rq
drivers//scsi/scsi_lib.c:1824:16: error: 'scsi_cleanup_rq' undeclared here (not in a function); did you mean 'scsi_queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~~~~~~
scsi_queue_rq
>> drivers//scsi/scsi_lib.c:1824:16: error: positional initialization of field in 'struct' declared with 'designated_init' attribute [-Werror=designated-init]
drivers//scsi/scsi_lib.c:1824:16: note: (near initialization for 'scsi_mq_ops')
cc1: some warnings being treated as errors
vim +1824 drivers//scsi/scsi_lib.c
1811
1812 static const struct blk_mq_ops scsi_mq_ops = {
1813 .get_budget = scsi_mq_get_budget,
1814 .put_budget = scsi_mq_put_budget,
1815 .queue_rq = scsi_queue_rq,
1816 .complete = scsi_softirq_done,
1817 .timeout = scsi_timeout,
1818 #ifdef CONFIG_BLK_DEBUG_FS
1819 .show_rq = scsi_show_rq,
1820 #endif
1821 .init_request = scsi_mq_init_request,
1822 .exit_request = scsi_mq_exit_request,
1823 .initialize_rq_fn = scsi_initialize_rq,
> 1824 .cleanup_rq = scsi_cleanup_rq,
1825 .busy = scsi_mq_lld_busy,
1826 .map_queues = scsi_map_queues,
1827 };
1828
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
PATCH 2/2 scsi: core: fix dh and multipathing for SCSI hosts without request batching
In reply to Steffen MaierThis was missing from scsi_device_from_queue() due to the introduction of another new scsi_mq_ops_no_commit of linux-next commit 8930a6c20791 ("scsi: core: add support for request batching") from Martin's scsi/5.4/scsi-queue or James' scsi/misc.
Only devicehandler code seems to call scsi_device_from_queue():
*** drivers/scsi/scsi_dh.c:
scsi_dh_activate[255] sdev = scsi_device_from_queue(q); scsi_dh_set_params[302] sdev = scsi_device_from_queue(q); scsi_dh_attach[325] sdev = scsi_device_from_queue(q);
scsi_dh_attached_handler_name[363] sdev = scsi_device_from_queue(q);
Fixes multipath tools follow-on errors:
$ multipath -v6 ... libdevmapper: ioctl/libdm-iface.c(1887): device-mapper: reload ioctl on mpatha failed: No such device ... mpatha: failed to load map, error 19 ...
showing also as kernel messages:
device-mapper: table: 252:0: multipath: error attaching hardware handler
device-mapper: ioctl: error adding target to table
--- drivers/scsi/scsi_lib.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/scsi/scsi_lib.c +2 -1
@@ -1922,7 +1922,8 @@ struct scsi_device *scsi_device_from_queue(struct request_queue *q) {struct scsi_device *sdev =NULL ; -if (q->mq_ops == &scsi_mq_ops) +if (q->mq_ops == &scsi_mq_ops_no_commit || + q->mq_ops == &scsi_mq_ops) sdev = q->queuedata;if (!sdev || !get_device (&sdev->sdev_gendev)) sdev =NULL ;
Signature · 1 line
2.17.1
kbuild test robot
build botRe: [PATCH 2/2] scsi: core: fix dh and multipathing for SCSI hosts without request batching
I love your patch! Yet something to improve:
Hi Steffen,
I love your patch! Yet something to improve:
[auto build test ERROR on linus/master]
[cannot apply to v5.3-rc3 next-20190807]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Steffen-Maier/scsi-core-fix-missing-cleanup_rq-for-SCSI-hosts-without-request-batching/20190808-052017
config: riscv-defconfig (attached as .config)
compiler: riscv64-linux-gcc (GCC) 7.4.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.4.0 make.cross ARCH=riscv
If you fix the issue, kindly add following tag
All errors (new ones prefixed by >>):
drivers/scsi/scsi_lib.c:1824:3: error: 'const struct blk_mq_ops' has no member named 'cleanup_rq'; did you mean 'queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~
queue_rq
drivers/scsi/scsi_lib.c:1824:16: error: 'scsi_cleanup_rq' undeclared here (not in a function); did you mean 'scsi_queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~~~~~~
scsi_queue_rq
drivers/scsi/scsi_lib.c: In function 'scsi_device_from_queue':
>> drivers/scsi/scsi_lib.c:1881:20: error: 'scsi_mq_ops_no_commit' undeclared (first use in this function); did you mean 'scsi_mq_ops'?
if (q->mq_ops == &scsi_mq_ops_no_commit ||
^~~~~~~~~~~~~~~~~~~~~
scsi_mq_ops drivers/scsi/scsi_lib.c:1881:20: note: each undeclared identifier is reported only once for each function it appears in
vim +1881 drivers/scsi/scsi_lib.c
1811
1812 static const struct blk_mq_ops scsi_mq_ops = {
1813 .get_budget = scsi_mq_get_budget,
1814 .put_budget = scsi_mq_put_budget,
1815 .queue_rq = scsi_queue_rq,
1816 .complete = scsi_softirq_done,
1817 .timeout = scsi_timeout,
1818 #ifdef CONFIG_BLK_DEBUG_FS
1819 .show_rq = scsi_show_rq,
1820 #endif
1821 .init_request = scsi_mq_init_request,
1822 .exit_request = scsi_mq_exit_request,
1823 .initialize_rq_fn = scsi_initialize_rq,
> 1824 .cleanup_rq = scsi_cleanup_rq,
1825 .busy = scsi_mq_lld_busy,
1826 .map_queues = scsi_map_queues,
1827 };
1828
1829 struct request_queue *scsi_mq_alloc_queue(struct scsi_device *sdev)
1830 {
1831 sdev->request_queue = blk_mq_init_queue(&sdev->host->tag_set);
1832 if (IS_ERR(sdev->request_queue))
1833 return NULL;
1834
1835 sdev->request_queue->queuedata = sdev;
1836 __scsi_init_queue(sdev->host, sdev->request_queue);
1837 blk_queue_flag_set(QUEUE_FLAG_SCSI_PASSTHROUGH, sdev->request_queue);
1838 return sdev->request_queue;
1839 }
1840
1841 int scsi_mq_setup_tags(struct Scsi_Host *shost)
1842 {
1843 unsigned int cmd_size, sgl_size;
1844
1845 sgl_size = scsi_mq_inline_sgl_size(shost);
1846 cmd_size = sizeof(struct scsi_cmnd) + shost->hostt->cmd_size + sgl_size;
1847 if (scsi_host_get_prot(shost))
1848 cmd_size += sizeof(struct scsi_data_buffer) +
1849 sizeof(struct scatterlist) * SCSI_INLINE_PROT_SG_CNT;
1850
1851 memset(&shost->tag_set, 0, sizeof(shost->tag_set));
1852 shost->tag_set.ops = &scsi_mq_ops;
1853 shost->tag_set.nr_hw_queues = shost->nr_hw_queues ? : 1;
1854 shost->tag_set.queue_depth = shost->can_queue;
1855 shost->tag_set.cmd_size = cmd_size;
1856 shost->tag_set.numa_node = NUMA_NO_NODE;
1857 shost->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
1858 shost->tag_set.flags |=
1859 BLK_ALLOC_POLICY_TO_MQ_FLAG(shost->hostt->tag_alloc_policy);
1860 shost->tag_set.driver_data = shost;
1861
1862 return blk_mq_alloc_tag_set(&shost->tag_set);
1863 }
1864
1865 void scsi_mq_destroy_tags(struct Scsi_Host *shost)
1866 {
1867 blk_mq_free_tag_set(&shost->tag_set);
1868 }
1869
1870 /**
1871 * scsi_device_from_queue - return sdev associated with a request_queue
1872 * @q: The request queue to return the sdev from
1873 *
1874 * Return the sdev associated with a request queue or NULL if the
1875 * request_queue does not reference a SCSI device.
1876 */
1877 struct scsi_device *scsi_device_from_queue(struct request_queue *q)
1878 {
1879 struct scsi_device *sdev = NULL;
1880
> 1881 if (q->mq_ops == &scsi_mq_ops_no_commit || 1882 q->mq_ops == &scsi_mq_ops)
1883 sdev = q->queuedata;
1884 if (!sdev || !get_device(&sdev->sdev_gendev))
1885 sdev = NULL;
1886
1887 return sdev;
1888 }
1889 EXPORT_SYMBOL_GPL(scsi_device_from_queue);
1890
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
kbuild test robot
build botRe: [PATCH 2/2] scsi: core: fix dh and multipathing for SCSI hosts without request batching
In reply to Steffen MaierI love your patch! Perhaps something to improve:
Hi Steffen,
I love your patch! Perhaps something to improve:
[auto build test WARNING on linus/master]
[cannot apply to v5.3-rc3 next-20190808]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Steffen-Maier/scsi-core-fix-missing-cleanup_rq-for-SCSI-hosts-without-request-batching/20190808-052017
config: i386-randconfig-d003-201931 (attached as .config)
compiler: gcc-7 (Debian 7.4.0-10) 7.4.0 reproduce:
# save the attached .config to linux build tree
make ARCH=i386 If you fix the issue, kindly add following tag
All warnings (new ones prefixed by >>):
drivers/scsi/scsi_lib.c:1824:3: error: 'const struct blk_mq_ops' has no member named 'cleanup_rq'; did you mean 'queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~
queue_rq
drivers/scsi/scsi_lib.c:1824:16: error: 'scsi_cleanup_rq' undeclared here (not in a function); did you mean 'scsi_queue_rq'?
.cleanup_rq = scsi_cleanup_rq,
^~~~~~~~~~~~~~~
scsi_queue_rq
In file included from include/linux/export.h:45:0,
from include/linux/linkage.h:7,
from include/linux/fs.h:5,
from include/linux/highmem.h:5,
from include/linux/bio.h:8,
from drivers/scsi/scsi_lib.c:12:
drivers/scsi/scsi_lib.c: In function 'scsi_device_from_queue':
drivers/scsi/scsi_lib.c:1881:20: error: 'scsi_mq_ops_no_commit' undeclared (first use in this function); did you mean 'scsi_mq_ops'?
if (q->mq_ops == &scsi_mq_ops_no_commit ||
^
include/linux/compiler.h:58:52: note: in definition of macro '__trace_if_var'
#define __trace_if_var(cond) (__builtin_constant_p(cond) ? (cond) : __trace_if_value(cond))
^~~~
>> drivers/scsi/scsi_lib.c:1881:2: note: in expansion of macro 'if'
if (q->mq_ops == &scsi_mq_ops_no_commit ||
^~
drivers/scsi/scsi_lib.c:1881:20: note: each undeclared identifier is reported only once for each function it appears in
if (q->mq_ops == &scsi_mq_ops_no_commit ||
^
include/linux/compiler.h:58:52: note: in definition of macro '__trace_if_var'
#define __trace_if_var(cond) (__builtin_constant_p(cond) ? (cond) : __trace_if_value(cond))
^~~~
>> drivers/scsi/scsi_lib.c:1881:2: note: in expansion of macro 'if'
if (q->mq_ops == &scsi_mq_ops_no_commit ||
^~vim +/if +1881 drivers/scsi/scsi_lib.c
1869
1870 /**
1871 * scsi_device_from_queue - return sdev associated with a request_queue
1872 * @q: The request queue to return the sdev from
1873 *
1874 * Return the sdev associated with a request queue or NULL if the
1875 * request_queue does not reference a SCSI device.
1876 */
1877 struct scsi_device *scsi_device_from_queue(struct request_queue *q)
1878 {
1879 struct scsi_device *sdev = NULL;
1880
> 1881 if (q->mq_ops == &scsi_mq_ops_no_commit ||
1882 q->mq_ops == &scsi_mq_ops)
1883 sdev = q->queuedata;
1884 if (!sdev || !get_device(&sdev->sdev_gendev))
1885 sdev = NULL;
1886
1887 return sdev;
1888 }
1889 EXPORT_SYMBOL_GPL(scsi_device_from_queue);
1890
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
Quoted 11 lines · Steffen Maier
Hi James, Martin, Paolo, Ming,
multipathing with linux-next is broken since 20190723 in our CI. The patches fix a memleak and a severe dh/multipath functional regression. It would be nice if we could get them to 5.4/scsi-queue and also next.
I would have preferred if such a new feature had used its own new copy
scsi_mq_ops_batchinginstead of changing the use case and semantics of the existingscsi_mq_ops, because this would likely cause less regressions for all the other users not using the new feature.
Quoted 9 lines
Steffen Maier (2):
scsi: core: fix missing .cleanup_rq for SCSI hosts without request
batching
scsi: core: fix dh and multipathing for SCSI hosts without request
batchingdrivers/scsi/scsi_lib.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 10 lines · Steffen Maier
Hi James, Martin, Paolo, Ming,
multipathing with linux-next is broken since 20190723 in our CI. The patches fix a memleak and a severe dh/multipath functional regression. It would be nice if we could get them to 5.4/scsi-queue and also next.
I would have preferred if such a new feature had used its own new copy
scsi_mq_ops_batchinginstead of changing the use case and semantics of the existingscsi_mq_ops, because this would likely cause less regressions for all the other users not using the new feature.
For both patches:
BTW, these two patches fix several nvmeof-mp blktests regressions.
Bart.