- 14 Dec, 2022 2 commits
-
-
Michael Kelley authored
storvsc_queuecommand() maps the scatter/gather list using scsi_dma_map(), which in a confidential VM allocates swiotlb bounce buffers. If the I/O submission fails in storvsc_do_io(), the I/O is typically retried by higher level code, but the bounce buffer memory is never freed. The mostly like cause of I/O submission failure is a full VMBus channel ring buffer, which is not uncommon under high I/O loads. Eventually enough bounce buffer memory leaks that the confidential VM can't do any I/O. The same problem can arise in a non-confidential VM with kernel boot parameter swiotlb=force. Fix this by doing scsi_dma_unmap() in the case of an I/O submission error, which frees the bounce buffer memory. Fixes: 743b237c ("scsi: storvsc: Add Isolation VM support for storvsc driver") Signed-off-by:
Michael Kelley <mikelley@microsoft.com> Link: https://lore.kernel.org/r/1670183564-76254-1-git-send-email-mikelley@microsoft.comTested-by:
Dexuan Cui <decui@microsoft.com> Reviewed-by:
Tianyu Lan <Tianyu.Lan@microsoft.com> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com>
-
Wenchao Hao authored
It was observed that the kernel would potentially send ISCSI_KEVENT_UNBIND_SESSION multiple times. Introduce 'target_state' in iscsi_cls_session() to make sure session will send only one unbind session event. This introduces a regression wrt. the issue fixed in commit 13e60d3b ("scsi: iscsi: Report unbind session event when the target has been removed"). If iscsid dies for any reason after sending an unbind session to kernel, once iscsid is restarted, the kernel's ISCSI_KEVENT_UNBIND_SESSION event is lost and userspace is then unable to logout. However, the session is actually in invalid state (its target_id is INVALID) so iscsid should not sync this session during restart. Consequently we need to check the session's target state during iscsid restart. If session is in unbound state, do not sync this session and perform session teardown. This is OK because once a session is unbound, we can not recover it any more (mainly because its target id is INVALID). Signed-off-by:
Wenchao Hao <haowenchao@huawei.com> Link: https://lore.kernel.org/r/20221126010752.231917-1-haowenchao@huawei.comReviewed-by:
Mike Christie <michael.christie@oracle.com> Reviewed-by:
Wu Bo <wubo40@huawei.com> Signed-off-by:
-
- 01 Dec, 2022 7 commits
-
-
Arun Easi authored
While performing CPU hotplug, a crash with the following stack was seen: Call Trace: qla24xx_process_response_queue+0x42a/0x970 [qla2xxx] qla2x00_start_nvme_mq+0x3a2/0x4b0 [qla2xxx] qla_nvme_post_cmd+0x166/0x240 [qla2xxx] nvme_fc_start_fcp_op.part.0+0x119/0x2e0 [nvme_fc] blk_mq_dispatch_rq_list+0x17b/0x610 __blk_mq_sched_dispatch_requests+0xb0/0x140 blk_mq_sched_dispatch_requests+0x30/0x60 __blk_mq_run_hw_queue+0x35/0x90 __blk_mq_delay_run_hw_queue+0x161/0x180 blk_execute_rq+0xbe/0x160 __nvme_submit_sync_cmd+0x16f/0x220 [nvme_core] nvmf_connect_admin_queue+0x11a/0x170 [nvme_fabrics] nvme_fc_create_association.cold+0x50/0x3dc [nvme_fc] nvme_fc_connect_ctrl_work+0x19/0x30 [nvme_fc] process_one_work+0x1e8/0x3c0 On abort timeout, completion was called without checking if the I/O was already completed. Verify that I/O and abort request are indeed outstanding before attempting completion. Fixes: 71c80b75 ("scsi: qla2xxx: Do command completion on abort timeout") Reported-by:Marco Patalano <mpatalan@redhat.com> Tested-by:
Arun Easi <aeasi@marvell.com> Signed-off-by:
Nilesh Javali <njavali@marvell.com> Link: https://lore.kernel.org/r/20221129092634.15347-1-njavali@marvell.comSigned-off-by:
-
Mike Christie authored
This converts the NVMe errors we commonly see during PR handling to PR_STS errors or -Exyz errors. pr_ops callers can then handle SCSI and NVMe errors without knowing the device types. Signed-off-by:
Chaitanya Kulkarni <kch@nvidia.com> Reviewed-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Mike Christie authored
This converts the SCSI errors we commonly see during PR handling to PR_STS errors or -Exyz errors. pr_ops callers can then handle SCSI and NVMe errors without knowing the device types. Signed-off-by:
-
Mike Christie authored
The next patch adds a helper status_byte function that works like host_byte, so this patch renames the old status_byte to sg_status_byte since it's only used for SG IO. Signed-off-by:
-
Mike Christie authored
If a PR operation fails we can return a device-specific error which is impossible to handle in some cases because we could have a mix of devices when DM is used, or future users like LIO only knows it's interacting with a block device so it doesn't know the type. This patch adds a new pr_status enum so drivers can convert errors to a common type which can be handled by the caller. Signed-off-by:
-
Johannes Thumshirn authored
Add tracepoints to the SCSI zone append emulation in order to trace the zone start to write-pointer aligned LBA translation and the corresponding completion. Signed-off-by:
Johannes Thumshirn <johannes.thumshirn@wdc.com> Link: https://lore.kernel.org/r/d103bcf5f90139143469f2a0084c74bd9e03ad4a.1669804487.git.johannes.thumshirn@wdc.comReviewed-by:
Jason Yan <yanaijie@huawei.com> Signed-off-by:
-
Christophe JAILLET authored
This file does not use rcu, so there is no point in including <linux/rculist.h>. The dependency has been removed in commit fa519f70 ("scsi: libfc: fixup 'sleeping function called from invalid context'") It turned a list_for_each_entry_rcu() into a list_for_each_entry(). So just #include <linux/list.h> now. Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Link: https://lore.kernel.org/r/960f34418358f0c35e645aa2cf7e0ec7fe6b60b9.1669461197.git.christophe.jaillet@wanadoo.frSigned-off-by:
-
- 26 Nov, 2022 23 commits
-
-
Kirill A. Shutemov authored
get_user() expects the pointer to be pointer-to-simple-variable type, but sic->data is array of 'unsigned char'. It violates get_user() contracts. Explicitly take pointer to the first element of the array. It matches current behaviour. This is preparation for fixing sparse warnings caused by Linear Address Masking patchset. Signed-off-by:
Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Link: https://lore.kernel.org/r/20221117232304.1544-1-kirill.shutemov@linux.intel.com Cc: "James E.J. Bottomley" <jejb@linux.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Signed-off-by:
-
Yu Zhe authored
Fix typos in comment. Reviewed-by:
Randy Dunlap <rdunlap@infradead.org> Signed-off-by:
Yu Zhe <yuzhe@nfschina.com> Link: https://lore.kernel.org/r/20221125020703.22216-1-yuzhe@nfschina.comSigned-off-by:
-
John Garry authored
Instead of using hardcoded '0' as the do_scsi_scan_host() -> scsi_scan_host_selected() rescan arg, use proper macro SCSI_SCAN_INITIAL. Signed-off-by:
John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20221121121725.1910795-3-john.g.garry@oracle.comReviewed-by:
Damien Le Moal <damien.lemoal@opensource.wdc.com> Reviewed-by:
Bart Van Assche <bvanassche@acm.org> Signed-off-by:
-
John Garry authored
Instead of using hardcoded '1' as the __scsi_add_device() -> scsi_probe_and_add_lun() rescan arg, use proper macro SCSI_SCAN_RESCAN. Signed-off-by:
-
ChanWoo Lee authored
Modify to remove unnecessary 'return 0' code. Signed-off-by:
ChanWoo Lee <cw9316.lee@samsung.com> Link: https://lore.kernel.org/r/20221121003338.11034-1-cw9316.lee@samsung.comReviewed-by:
-
Bart Van Assche authored
Fix the following issues in ufshcd_poll(): - If polling succeeds, return a positive value. - Do not complete polling requests from interrupt context because the block layer expects these requests to be completed from thread context. From block/bio.c: If REQ_ALLOC_CACHE is set, the final put of the bio MUST be done from process context, not hard/soft IRQ. Fixes: eaab9b57 ("scsi: ufs: Implement polling support") Signed-off-by:Adrian Hunter <adrian.hunter@intel.com> Signed-off-by:
-
Jie Zhan authored
sas_ata_wait_after_reset() does not need to be exported since it is no longer referenced outside libsas. Signed-off-by:
Jie Zhan <zhanjie9@hisilicon.com> Link: https://lore.kernel.org/r/20221118083714.4034612-6-zhanjie9@hisilicon.comReviewed-by:
John Garry <john.garry@huawei.com> Signed-off-by:
-
Jie Zhan authored
SATA devices on an expander may be removed and not be found again when I_T nexus reset and revalidation are processed simultaneously. The issue comes from: - Revalidation can remove SATA devices in link reset, e.g. in hisi_sas_clear_nexus_ha(). - However, hisi_sas_debug_I_T_nexus_reset() polls the state of a SATA device on an expander after sending link_reset, where it calls: hisi_sas_debug_I_T_nexus_reset sas_ata_wait_after_reset ata_wait_after_reset ata_wait_ready smp_ata_check_ready sas_ex_phy_discover sas_ex_phy_discover_helper sas_set_ex_phy The ex_phy's change count is updated in sas_set_ex_phy(), so SATA devices after a link reset may not be found later through revalidation. A similar issue was reported in: commit 0f3fce5c ("[SCSI] libsas: fix ata_eh clobbering ex_phys via smp_ata_check_ready") commit 87c8331f ("[SCSI] libsas: prevent domain rediscovery competing with ata error handling"). To address this issue, in hisi_sas_debug_I_T_nexus_reset(), we now call smp_ata_check_ready_type() that only polls the device type while not updating the ex_phy's data of libsas. Fixes: 71453bd9 ("scsi: hisi_sas: Use sas_ata_wait_after_reset() in IT nexus reset") Signed-off-by: -
Jie Zhan authored
Create function smp_ata_check_ready_type() for LLDDs to wait for SATA devices to come up after a link reset. Signed-off-by:
-
Jie Zhan authored
This reverts commit f5f2a271. This is now unnecessary to solve the SATA devices missing issue in hisi_sas_clear_nexus_ha(). Hence, we should not ignore bcast events during sas_eh_handle_sas_errors() in case of missing bcast events, unless a justified need is found and a mechanism to defer (but not ignore) bcast events in sas_eh_handle_sas_errors() is provided. Also, in hisi_sas_clear_nexus_ha(), there is nothing further to handle in "out: " other than return, so that part can be reverted. Signed-off-by:
-
Jie Zhan authored
This reverts commit 11ff0c98. Draining or flushing events in hisi_sas_rescan_topology() can hang the driver, typically with phy up or phy down events being processed, i.e. sas_porte_bytes_dmaed() or sas_phye_loss_of_signal(). Signed-off-by:
-
ChanWoo Lee authored
Be consistent with the rest of driver wrt. functions returning bool. 91: return !!(host->caps & UFS_MTK_CAP_BOOST_CRYPT_ENGINE); 98: return !!(host->caps & UFS_MTK_CAP_VA09_PWR_CTRL); 105: return !!(host->caps & UFS_MTK_CAP_BROKEN_VCC); Signed-off-by:
Stanley Chu <stanley.chu@mediatek.com> Reviewed-by:
AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by:
-
ChanWoo Lee authored
Remove unnecessary if/goto code. Signed-off-by:
-
Bart Van Assche authored
Since commit f93ed747 ("scsi: core: Release SCSI devices synchronously"), scsi_device_put() might sleep. Avoid calling it from alua_rtpg_queue() with the pg_lock held. The lock only pretects h->pg, anyway. To avoid the pg being freed under us, because of a race with another thread, take a temporary reference. In alua_rtpg_queue(), verify that the pg still belongs to the sdev being passed before actually queueing the RTPG. This patch fixes the following smatch warning: drivers/scsi/device_handler/scsi_dh_alua.c:1013 alua_rtpg_queue() warn: sleeping in atomic context alua_check_vpd() <- disables preempt -> alua_rtpg_queue() -> scsi_device_put() Cc: Martin Wilck <mwilck@suse.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Sachin Sant <sachinp@linux.ibm.com> Cc: Benjamin Block <bblock@linux.ibm.com> Suggested-by:
Martin Wilck <mwilck@suse.com> Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
Sachin Sant <sachinp@linux.ibm.com> Signed-off-by:
-
Bart Van Assche authored
There is a bug in commit 0b25e17e ("scsi: alua: Move a scsi_device_put() call out of alua_check_vpd()"): that patch may cause alua_rtpg_queue() callers to call scsi_device_put() even if that function should not be called. Revert that commit to prepare for a different solution. Cc: Hannes Reinecke <hare@suse.de> Cc: Martin Wilck <mwilck@suse.com> Cc: Sachin Sant <sachinp@linux.ibm.com> Cc: Benjamin Block <bblock@linux.ibm.com> Reported-by:
Benjamin Block <bblock@linux.ibm.com> Signed-off-by:
-
Gaosheng Cui authored
Smatch reports a warning as follows: drivers/scsi/snic/snic_disc.c:307 snic_tgt_create() warn: '&tgt->list' not removed from list If device_add() fails in snic_tgt_create(), tgt will be freed, but tgt->list will not be removed from snic->disc.tgt_list, then list traversal may cause UAF. Remove from snic->disc.tgt_list before free(). Fixes: c8806b6c ("snic: driver for Cisco SCSI HBA") Signed-off-by:
Gaosheng Cui <cuigaosheng1@huawei.com> Link: https://lore.kernel.org/r/20221117035100.2944812-1-cuigaosheng1@huawei.comAcked-by:
Narsimhulu Musini <nmusini@cisco.com> Signed-off-by:
-
Gleb Chesnokov authored
Initialization of vha->unknown_atio_list and vha->unknown_atio_work only happens for base_vha in qlt_probe_one_stage1(). But there is no initialization for NPIV hosts that are created in qla24xx_vport_create(). This causes a crash when trying to access these NPIV host fields. Fix this by adding initialization to qla_vport_create(). Signed-off-by:
Gleb Chesnokov <gleb.chesnokov@scst.dev> Link: https://lore.kernel.org/r/376c89a2-a9ac-bcf9-bf0f-dfe89a02fd4b@scst.devSigned-off-by:
-
Gleb Chesnokov authored
Commit 9b3e0f4d ("scsi: qla2xxx: Move work element processing out of DPC thread") introduced the initialization of vha->iocb_work in qla2x00_create_host() function. This initialization is also called from qla2x00_probe_one() function, just after qla2x00_create_host(). Hence remove this duplicate call since it has already been called before. Signed-off-by:
-
Chen Zhongjin authored
fcoe_init() calls fcoe_transport_attach(&fcoe_sw_transport), but when fcoe_if_init() fails, &fcoe_sw_transport is not detached and leaves freed &fcoe_sw_transport on fcoe_transports list. This causes panic when reinserting module. BUG: unable to handle page fault for address: fffffbfff82e2213 RIP: 0010:fcoe_transport_attach+0xe1/0x230 [libfcoe] Call Trace: <TASK> do_one_initcall+0xd0/0x4e0 load_module+0x5eee/0x7210 ... Fixes: 78a58246 ("[SCSI] fcoe: convert fcoe.ko to become an fcoe transport provider driver") Signed-off-by:
Chen Zhongjin <chenzhongjin@huawei.com> Link: https://lore.kernel.org/r/20221115092442.133088-1-chenzhongjin@huawei.comSigned-off-by:
-
Shin'ichiro Kawasaki authored
ZBC Zoned Block Commands specification mandates SYNCHRONIZE CACHE(16) for host-managed zoned block devices, but does not mandate SYNCHRONIZE CACHE(10). Call SYNCHRONIZE CACHE(16) in place of SYNCHRONIZE CACHE(10) to ensure that the command is always supported. For this purpose, add use_16_for_sync flag to struct scsi_device in same manner as use_16_for_rw flag. To be precise, ZBC does not mandate SYNCHRONIZE CACHE(16) for host-aware zoned block devices. However, modern devices should support 16-byte commands. Hence, call SYNCHRONIZE CACHE (16) on both types of ZBC devices, host-aware and host-managed. Of note is that READ(16) and WRITE(16) have same story and they are already called for both types of ZBC devices. Another note is that this patch depends on the fix commit ea045fd3 ("ata: libata-scsi: fix SYNCHRONIZE CACHE (16) command failure"). Signed-off-by:
Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Link: https://lore.kernel.org/r/20221115002905.1709006-1-shinichiro.kawasaki@wdc.comReviewed-by:
Damien Le Moal <damien.lemoal@opendource.wdc.com> Signed-off-by:
-
Shang XiaoJing authored
ipr_init() will not call unregister_reboot_notifier() when pci_register_driver() fails, which causes a WARNING. Call unregister_reboot_notifier() when pci_register_driver() fails. notifier callback ipr_halt [ipr] already registered WARNING: CPU: 3 PID: 299 at kernel/notifier.c:29 notifier_chain_register+0x16d/0x230 Modules linked in: ipr(+) xhci_pci_renesas xhci_hcd ehci_hcd usbcore led_class gpu_sched drm_buddy video wmi drm_ttm_helper ttm drm_display_helper drm_kms_helper drm drm_panel_orientation_quirks agpgart cfbft CPU: 3 PID: 299 Comm: modprobe Tainted: G W 6.1.0-rc1-00190-g39508d23b672-dirty #332 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014 RIP: 0010:notifier_chain_register+0x16d/0x230 Call Trace: <TASK> __blocking_notifier_chain_register+0x73/0xb0 ipr_init+0x30/0x1000 [ipr] do_one_initcall+0xdb/0x480 do_init_module+0x1cf/0x680 load_module+0x6a50/0x70a0 __do_sys_finit_module+0x12f/0x1c0 do_syscall_64+0x3f/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd Fixes: f72919ec ("[SCSI] ipr: implement shutdown changes and remove obsolete write cache parameter") Signed-off-by:
Shang XiaoJing <shangxiaojing@huawei.com> Link: https://lore.kernel.org/r/20221113064513.14028-1-shangxiaojing@huawei.comSigned-off-by:
-
Yang Yingliang authored
Afer commit 1fa5ae85 ("driver core: get rid of struct device's bus_id string array"), the name of device is allocated dynamically, it needs be freed when device_register() returns error. As comment of device_register() says, one should use put_device() to give up the reference in the error path. Fix this by calling put_device(), then the name can be freed in kobject_cleanup(), and sdbg_host is freed in sdebug_release_adapter(). When the device release is not set, it means the device is not initialized. We can not call put_device() in this case. Use kfree() to free memory. Fixes: 1fa5ae85 ("driver core: get rid of struct device's bus_id string array") Signed-off-by:
Yang Yingliang <yangyingliang@huawei.com> Link: https://lore.kernel.org/r/20221112131010.3757845-1-yangyingliang@huawei.comSigned-off-by:
-
Yang Yingliang authored
If device_register() returns an error, the name allocated by dev_set_name() needs to be freed. As the comment of device_register() says, one should use put_device() to give up the reference in the error path. Fix this by calling put_device(), then the name can be freed in kobject_cleanup(). The 'fcf' is freed in fcoe_fcf_device_release(), so the kfree() in the error path can be removed. The 'ctlr' is freed in fcoe_ctlr_device_release(), so don't use the error label, just return NULL after calling put_device(). Fixes: 9a74e884 ("[SCSI] libfcoe: Add fcoe_sysfs") Signed-off-by:
-
- 25 Nov, 2022 8 commits
-
-
Harshit Mogalapalli authored
As 'alloc_len' is user controlled data, if user tries to allocate memory larger than(>=) MAX_ORDER, then kcalloc() will fail, it creates a stack trace and messes up dmesg with a warning. Add __GFP_NOWARN in order to avoid too large allocation warning. This is detected by static analysis using smatch. Fixes: 7db0e0c8 ("scsi: scsi_debug: Fix buffer size of REPORT ZONES command") Signed-off-by:
Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Link: https://lore.kernel.org/r/20221112070612.2121535-1-harshit.m.mogalapalli@oracle.comSigned-off-by:
-
Harshit Mogalapalli authored
As 'vnum' is controlled by user, so if user tries to allocate memory larger than(>=) MAX_ORDER, then kcalloc() will fail, it creates a stack trace and messes up dmesg with a warning. Add __GFP_NOWARN in order to avoid too large allocation warning. This is detected by static analysis using smatch. Fixes: c3e2fe92 ("scsi: scsi_debug: Implement VERIFY(10), add VERIFY(16)") Signed-off-by:
-
Chen Zhongjin authored
In efct_device_init(), when efct_scsi_reg_fc_transport() fails, efct_scsi_tgt_driver_exit() is not called to release memory for efct_scsi_tgt_driver_init() and causes memleak: unreferenced object 0xffff8881020ce000 (size 2048): comm "modprobe", pid 465, jiffies 4294928222 (age 55.872s) backtrace: [<0000000021a1ef1b>] kmalloc_trace+0x27/0x110 [<000000004c3ed51c>] target_register_template+0x4fd/0x7b0 [target_core_mod] [<00000000f3393296>] efct_scsi_tgt_driver_init+0x18/0x50 [efct] [<00000000115de533>] 0xffffffffc0d90011 [<00000000d608f646>] do_one_initcall+0xd0/0x4e0 [<0000000067828cf1>] do_init_module+0x1cc/0x6a0 ... Fixes: 4df84e84 ("scsi: elx: efct: Driver initialization routines") Signed-off-by: -
ChanWoo Lee authored
Setting bitmap_len is not required when returning early. Defer until it is needed. Signed-off-by:
-
ChanWoo Lee authored
Only checks true and false so it can be converted to bool. Signed-off-by:
-
ChanWoo Lee authored
Separate the function name and message to make it easier to check the log. Modify messages to fit the format of others. Signed-off-by:
-
Yang Yingliang authored
If hpsa_sas_port_add_rphy() returns an error, the 'rphy' allocated in sas_end_device_alloc() needs to be freed. Address this by calling sas_rphy_free() in the error path. Fixes: d04e62b9 ("hpsa: add in sas transport class") Signed-off-by:
-
Yang Yingliang authored
hpsa_sas_port_add_phy() does: ... sas_phy_add() -> may return error here sas_port_add_phy() ... Whereas hpsa_free_sas_phy() does: ... sas_port_delete_phy() sas_phy_delete() ... If hpsa_sas_port_add_phy() returns an error, hpsa_free_sas_phy() can not be called to free the memory because the port and the phy have not been added yet. Replace hpsa_free_sas_phy() with sas_phy_free() and kfree() to avoid kernel crash in this case. Fixes: d04e62b9 ("hpsa: add in sas transport class") Signed-off-by:
-
