kernel/linux.git/drivers/ata, branch linux-6.0.y

ata: ahci: Fix PCS quirk application for suspend

2023-01-04T10:26:24+00:00

[ Upstream commit 37e14e4f3715428b809e4df9a9958baa64c77d51 ] Since kernel 5.3.4 my laptop (ICH8M controller) does not see Kingston SV300S37A60G SSD disk connected into a SATA connector on wake from suspend. The problem was introduced in c312ef176399 ("libata/ahci: Drop PCS quirk for Denverton and beyond"): the quirk is not applied on wake from suspend as it originally was. It is worth to mention the commit contained another bug: the quirk is not applied at all to controllers which require it. The fix commit 09d6ac8dc51a ("libata/ahci: Fix PCS quirk application") landed in 5.3.8. So testing my patch anywhere between commits c312ef176399 and 09d6ac8dc51a is pointless. Not all disks trigger the problem. For example nothing bad happens with Western Digital WD5000LPCX HDD. Test hardware: - Acer 5920G with ICH8M SATA controller - sda: some SATA HDD connnected into the DVD drive IDE port with a SATA-IDE caddy. It is a boot disk - sdb: Kingston SV300S37A60G SSD connected into the only SATA port Sample "dmesg --notime | grep -E '^(sd |ata)'" output on wake: sd 0:0:0:0: [sda] Starting disk sd 2:0:0:0: [sdb] Starting disk ata4: SATA link down (SStatus 4 SControl 300) ata3: SATA link down (SStatus 4 SControl 300) ata1.00: ACPI cmd ef/03:0c:00:00:00:a0 (SET FEATURES) filtered out ata1.00: ACPI cmd ef/03:42:00:00:00:a0 (SET FEATURES) filtered out ata1: FORCE: cable set to 80c ata5: SATA link down (SStatus 0 SControl 300) ata3: SATA link down (SStatus 4 SControl 300) ata3: SATA link down (SStatus 4 SControl 300) ata3.00: disabled sd 2:0:0:0: rejecting I/O to offline device ata3.00: detaching (SCSI 2:0:0:0) sd 2:0:0:0: [sdb] Start/Stop Unit failed: Result: hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK sd 2:0:0:0: [sdb] Synchronizing SCSI cache sd 2:0:0:0: [sdb] Synchronize Cache(10) failed: Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK sd 2:0:0:0: [sdb] Stopping disk sd 2:0:0:0: [sdb] Start/Stop Unit failed: Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK Commit c312ef176399 dropped ahci_pci_reset_controller() which internally calls ahci_reset_controller() and applies the PCS quirk if needed after that. It was called each time a reset was required instead of just ahci_reset_controller(). This patch puts the function back in place. Fixes: c312ef176399 ("libata/ahci: Drop PCS quirk for Denverton and beyond") Signed-off-by: Adam Vodopjan Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata: fix NCQ autosense logic

2022-12-31T12:25:50+00:00

[ Upstream commit 7390896b3484d44cbdb8bc4859964314ac66d3c9 ] Currently, the logic if we should call ata_scsi_set_sense() (and set flag ATA_QCFLAG_SENSE_VALID to indicate that we have successfully added sense data to the struct ata_queued_cmd) looks like this: if (dev->class == ATA_DEV_ZAC && ((qc->result_tf.status & ATA_SENSE) || qc->result_tf.auxiliary)) The problem with this is that a drive can support the NCQ command error log without supporting NCQ autosense. On such a drive, if the failing command has sense data, the status field in the NCQ command error log will have the ATA_SENSE bit set. It is just that this sense data is not included in the NCQ command error log when NCQ autosense is not supported. Instead the sense data has to be fetched using the REQUEST SENSE DATA EXT command. Therefore, we should only add the sense data if the drive supports NCQ autosense AND the ATA_SENSE bit is set in the status field. Fix this, and at the same time, remove the duplicated ATA_DEV_ZAC check. The struct ata_taskfile supplied to ata_eh_read_log_10h() is memset:ed before calling the function, so simply checking if qc->result_tf.auxiliary is set is sufficient to tell us that the log actually contained sense data. Fixes: d238ffd59d3c ("libata: do not attempt to retrieve sense code twice") Signed-off-by: Niklas Cassel Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-core: do not issue non-internal commands once EH is pending

2022-11-26T08:27:33+00:00

[ Upstream commit e20e81a24a4d58744a29715aac2f795cd1651955 ] While the ATA specification states that a device should return command aborted for all commands queued after the device has entered error state, since ATA only keeps the sense data for the latest command (in non-NCQ case), we really don't want to send block layer commands to the device after it has entered error state. (Only ATA EH commands should be sent, to read the sense data etc.) Currently, scsi_queue_rq() will check if scsi_host_in_recovery() (state is SHOST_RECOVERY), and if so, it will _not_ issue a command via: scsi_dispatch_cmd() -> host->hostt->queuecommand() (ata_scsi_queuecmd()) -> __ata_scsi_queuecmd() -> ata_scsi_translate() -> ata_qc_issue() Before commit e494f6a72839 ("[SCSI] improved eh timeout handler"), when receiving a TFES error IRQ, the call chain looked like this: ahci_error_intr() -> ata_port_abort() -> ata_do_link_abort() -> ata_qc_complete() -> ata_qc_schedule_eh() -> blk_abort_request() -> blk_rq_timed_out() -> q->rq_timed_out_fn() (scsi_times_out()) -> scsi_eh_scmd_add() -> scsi_host_set_state(shost, SHOST_RECOVERY) Which meant that as soon as an error IRQ was serviced, SHOST_RECOVERY would be set. However, after commit e494f6a72839 ("[SCSI] improved eh timeout handler"), scsi_times_out() will instead call scsi_abort_command() which will queue delayed work, and the worker function scmd_eh_abort_handler() will call scsi_eh_scmd_add(), which calls scsi_host_set_state(shost, SHOST_RECOVERY). So now, after the TFES error IRQ has been serviced, we need to wait for the SCSI workqueue to run its work before SHOST_RECOVERY gets set. It is worth noting that, even before commit e494f6a72839 ("[SCSI] improved eh timeout handler"), we could receive an error IRQ from the time when scsi_queue_rq() checks scsi_host_in_recovery(), to the time when ata_scsi_queuecmd() is actually called. In order to handle both the delayed setting of SHOST_RECOVERY and the window where we can receive an error IRQ, add a check against ATA_PFLAG_EH_PENDING (which gets set when servicing the error IRQ), inside ata_scsi_queuecmd() itself, while holding the ap->lock. (Since the ap->lock is held while servicing IRQs.) Fixes: e494f6a72839 ("[SCSI] improved eh timeout handler") Signed-off-by: Niklas Cassel Tested-by: John Garry Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-transport: fix error handling in ata_tdev_add()

2022-11-26T08:27:32+00:00

[ Upstream commit 1ff36351309e3eadcff297480baf4785e726de9b ] In ata_tdev_add(), the return value of transport_add_device() is not checked. As a result, it causes null-ptr-deref while removing the module, because transport_remove_device() is called to remove the device that was not added. Unable to handle kernel NULL pointer dereference at virtual address 00000000000000d0 CPU: 13 PID: 13603 Comm: rmmod Kdump: loaded Tainted: G W 6.1.0-rc3+ #36 pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : device_del+0x48/0x3a0 lr : device_del+0x44/0x3a0 Call trace: device_del+0x48/0x3a0 attribute_container_class_device_del+0x28/0x40 transport_remove_classdev+0x60/0x7c attribute_container_device_trigger+0x118/0x120 transport_remove_device+0x20/0x30 ata_tdev_delete+0x24/0x50 [libata] ata_tlink_delete+0x40/0xa0 [libata] ata_tport_delete+0x2c/0x60 [libata] ata_port_detach+0x148/0x1b0 [libata] ata_pci_remove_one+0x50/0x80 [libata] ahci_remove_one+0x4c/0x8c [ahci] Fix this by checking and handling return value of transport_add_device() in ata_tdev_add(). In the error path, device_del() is called to delete the device which was added earlier in this function, and ata_tdev_free() is called to free ata_dev. Fixes: d9027470b886 ("[libata] Add ATA transport class") Signed-off-by: Yang Yingliang Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-transport: fix error handling in ata_tlink_add()

2022-11-26T08:27:32+00:00

[ Upstream commit cf0816f6322c5c37ee52655f928e91ecf32da103 ] In ata_tlink_add(), the return value of transport_add_device() is not checked. As a result, it causes null-ptr-deref while removing the module, because transport_remove_device() is called to remove the device that was not added. Unable to handle kernel NULL pointer dereference at virtual address 00000000000000d0 CPU: 33 PID: 13850 Comm: rmmod Kdump: loaded Tainted: G W 6.1.0-rc3+ #12 pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : device_del+0x48/0x39c lr : device_del+0x44/0x39c Call trace: device_del+0x48/0x39c attribute_container_class_device_del+0x28/0x40 transport_remove_classdev+0x60/0x7c attribute_container_device_trigger+0x118/0x120 transport_remove_device+0x20/0x30 ata_tlink_delete+0x88/0xb0 [libata] ata_tport_delete+0x2c/0x60 [libata] ata_port_detach+0x148/0x1b0 [libata] ata_pci_remove_one+0x50/0x80 [libata] ahci_remove_one+0x4c/0x8c [ahci] Fix this by checking and handling return value of transport_add_device() in ata_tlink_add(). Fixes: d9027470b886 ("[libata] Add ATA transport class") Signed-off-by: Yang Yingliang Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-transport: fix error handling in ata_tport_add()

2022-11-26T08:27:32+00:00

[ Upstream commit 3613dbe3909dcc637fe6be00e4dc43b4aa0470ee ] In ata_tport_add(), the return value of transport_add_device() is not checked. As a result, it causes null-ptr-deref while removing the module, because transport_remove_device() is called to remove the device that was not added. Unable to handle kernel NULL pointer dereference at virtual address 00000000000000d0 CPU: 12 PID: 13605 Comm: rmmod Kdump: loaded Tainted: G W 6.1.0-rc3+ #8 pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : device_del+0x48/0x39c lr : device_del+0x44/0x39c Call trace: device_del+0x48/0x39c attribute_container_class_device_del+0x28/0x40 transport_remove_classdev+0x60/0x7c attribute_container_device_trigger+0x118/0x120 transport_remove_device+0x20/0x30 ata_tport_delete+0x34/0x60 [libata] ata_port_detach+0x148/0x1b0 [libata] ata_pci_remove_one+0x50/0x80 [libata] ahci_remove_one+0x4c/0x8c [ahci] Fix this by checking and handling return value of transport_add_device() in ata_tport_add(). Fixes: d9027470b886 ("[libata] Add ATA transport class") Signed-off-by: Yang Yingliang Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-transport: fix double ata_host_put() in ata_tport_add()

2022-11-26T08:27:32+00:00

[ Upstream commit 8c76310740807ade5ecdab5888f70ecb6d35732e ] In the error path in ata_tport_add(), when calling put_device(), ata_tport_release() is called, it will put the refcount of 'ap->host'. And then ata_host_put() is called again, the refcount is decreased to 0, ata_host_release() is called, all ports are freed and set to null. When unbinding the device after failure, ata_host_stop() is called to release the resources, it leads a null-ptr-deref(), because all the ports all freed and null. Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008 CPU: 7 PID: 18671 Comm: modprobe Kdump: loaded Tainted: G E 6.1.0-rc3+ #8 pstate: 80400009 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : ata_host_stop+0x3c/0x84 [libata] lr : release_nodes+0x64/0xd0 Call trace: ata_host_stop+0x3c/0x84 [libata] release_nodes+0x64/0xd0 devres_release_all+0xbc/0x1b0 device_unbind_cleanup+0x20/0x70 really_probe+0x158/0x320 __driver_probe_device+0x84/0x120 driver_probe_device+0x44/0x120 __driver_attach+0xb4/0x220 bus_for_each_dev+0x78/0xdc driver_attach+0x2c/0x40 bus_add_driver+0x184/0x240 driver_register+0x80/0x13c __pci_register_driver+0x4c/0x60 ahci_pci_driver_init+0x30/0x1000 [ahci] Fix this by removing redundant ata_host_put() in the error path. Fixes: 2623c7a5f279 ("libata: add refcounting to ata_host") Signed-off-by: Yang Yingliang Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: libata-scsi: fix SYNCHRONIZE CACHE (16) command failure

2022-11-16T09:04:06+00:00

commit ea045fd344cb15c164e9ffc8b8cffb6883df8475 upstream. SAT SCSI/ATA Translation specification requires SCSI SYNCHRONIZE CACHE (10) and (16) commands both shall be translated to ATA flush command. Also, ZBC Zoned Block Commands specification mandates SYNCHRONIZE CACHE (16) command support. However, libata translates only SYNCHRONIZE CACHE (10). This results in SYNCHRONIZE CACHE (16) command failures on SATA drives and then libata translation does not conform to ZBC. To avoid the failure, add support for SYNCHRONIZE CACHE (16). Signed-off-by: Shin'ichiro Kawasaki Cc: stable@vger.kernel.org Reviewed-by: Christoph Hellwig Reviewed-by: Martin K. Petersen Signed-off-by: Damien Le Moal Signed-off-by: Greg Kroah-Hartman

ata: palmld: fix return value check in palmld_pata_probe()

2022-11-10T17:17:17+00:00

[ Upstream commit 015618c3ec19584c83ff179fa631be8cec906aaf ] If devm_platform_ioremap_resource() fails, it never return NULL pointer, replace the check with IS_ERR(). Fixes: 57bf0f5a162d ("ARM: pxa: use pdev resource for palmld mmio") Signed-off-by: Yang Yingliang Reviewed-by: Sergey Shtylyov Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin

ata: pata_legacy: fix pdc20230_set_piomode()

2022-11-10T17:17:17+00:00

[ Upstream commit 171a93182eccd6e6835d2c86b40787f9f832efaa ] Clang gives a warning when compiling pata_legacy.c with 'make W=1' about the 'rt' local variable in pdc20230_set_piomode() being set but unused. Quite obviously, there is an outb() call missing to write back the updated variable. Moreover, checking the docs by Petr Soucek revealed that bitwise AND should have been done with a negated timing mask and the master/slave timing masks were swapped while updating... Fixes: 669a5db411d8 ("[libata] Add a bunch of PATA drivers.") Reported-by: Damien Le Moal Signed-off-by: Sergey Shtylyov Signed-off-by: Damien Le Moal Signed-off-by: Sasha Levin