| Message ID | 20260814035846.2032778-1-yangxingui@huawei.com |
|---|---|
| State | New |
| Headers | show |
| Series | [v1] ata: libata-eh: Add timeout table entry for STANDBY IMMEDIATE | expand |
On Fri, Aug 14, 2026 at 11:58:46AM +0800, Xingui Yang wrote: > ATA_CMD_STANDBYNOW1 is not in the ata_eh_cmd_timeout_table and falls > back to the 5s default (ATA_EH_CMD_DFL_TIMEOUT), which is too short > for some drives to complete a STANDBY IMMEDIATE command issued from > ata_dev_power_set_standby() via ata_exec_internal(): > > ata1.00: qc timeout after 5000 msecs (cmd 0xe0) > ata1.00: STANDBY IMMEDIATE failed (err_mask=0x4) > > Reuse the existing ata_eh_flush_timeouts for this command, as a drive > entering standby may need to flush its internal cache, making the flush > timeout a natural fit. > > Signed-off-by: Xingui Yang <yangxingui@huawei.com> > --- > drivers/ata/libata-eh.c | 2 ++ > include/linux/libata.h | 2 +- > 2 files changed, 3 insertions(+), 1 deletion(-) > > diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c > index 05df7ea6954a..8f9e12f460c3 100644 > --- a/drivers/ata/libata-eh.c > +++ b/drivers/ata/libata-eh.c > @@ -149,6 +149,8 @@ ata_eh_cmd_timeout_table[ATA_EH_CMD_TIMEOUT_TABLE_SIZE] = { > .timeouts = ata_eh_flush_timeouts }, > { .commands = CMDS(ATA_CMD_VERIFY), > .timeouts = ata_eh_reset_timeouts }, > + { .commands = CMDS(ATA_CMD_STANDBYNOW1), > + .timeouts = ata_eh_flush_timeouts }, > }; > #undef CMDS > > diff --git a/include/linux/libata.h b/include/linux/libata.h > index 96e626d6a7ca..6c0d385b2cde 100644 > --- a/include/linux/libata.h > +++ b/include/linux/libata.h > @@ -425,7 +425,7 @@ enum { > /* This should match the actual table size of > * ata_eh_cmd_timeout_table in libata-eh.c. > */ > - ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 8, > + ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 9, > > /* User visible DMA mask for DMA control. DO NOT renumber. */ > ATA_DMA_MASK_ATA = (1 << 0), /* DMA on ATA Disk */ > -- > 2.43.0 > This patch does not seem to be on top of latest Linus tree. If you look at latest Linus tree, it already has: 1e024d2b41ee ("ata: libata-eh: Increase STANDBY IMMEDIATE timeout") Which looks to solve the same problem as this patch. Kind regards, Niklas
Hi, Niklas. On 2026/8/14 19:05, Niklas Cassel wrote: > On Fri, Aug 14, 2026 at 11:58:46AM +0800, Xingui Yang wrote: >> ATA_CMD_STANDBYNOW1 is not in the ata_eh_cmd_timeout_table and falls >> back to the 5s default (ATA_EH_CMD_DFL_TIMEOUT), which is too short >> for some drives to complete a STANDBY IMMEDIATE command issued from >> ata_dev_power_set_standby() via ata_exec_internal(): >> >> ata1.00: qc timeout after 5000 msecs (cmd 0xe0) >> ata1.00: STANDBY IMMEDIATE failed (err_mask=0x4) >> >> Reuse the existing ata_eh_flush_timeouts for this command, as a drive >> entering standby may need to flush its internal cache, making the flush >> timeout a natural fit. >> >> Signed-off-by: Xingui Yang <yangxingui@huawei.com> >> --- >> drivers/ata/libata-eh.c | 2 ++ >> include/linux/libata.h | 2 +- >> 2 files changed, 3 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c >> index 05df7ea6954a..8f9e12f460c3 100644 >> --- a/drivers/ata/libata-eh.c >> +++ b/drivers/ata/libata-eh.c >> @@ -149,6 +149,8 @@ ata_eh_cmd_timeout_table[ATA_EH_CMD_TIMEOUT_TABLE_SIZE] = { >> .timeouts = ata_eh_flush_timeouts }, >> { .commands = CMDS(ATA_CMD_VERIFY), >> .timeouts = ata_eh_reset_timeouts }, >> + { .commands = CMDS(ATA_CMD_STANDBYNOW1), >> + .timeouts = ata_eh_flush_timeouts }, >> }; >> #undef CMDS >> >> diff --git a/include/linux/libata.h b/include/linux/libata.h >> index 96e626d6a7ca..6c0d385b2cde 100644 >> --- a/include/linux/libata.h >> +++ b/include/linux/libata.h >> @@ -425,7 +425,7 @@ enum { >> /* This should match the actual table size of >> * ata_eh_cmd_timeout_table in libata-eh.c. >> */ >> - ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 8, >> + ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 9, >> >> /* User visible DMA mask for DMA control. DO NOT renumber. */ >> ATA_DMA_MASK_ATA = (1 << 0), /* DMA on ATA Disk */ >> -- >> 2.43.0 >> > > This patch does not seem to be on top of latest Linus tree. > > If you look at latest Linus tree, it already has: > 1e024d2b41ee ("ata: libata-eh: Increase STANDBY IMMEDIATE timeout") > > Which looks to solve the same problem as this patch. Aha, thanks for the quick review! I usually base my work on the "for-next" branches of "kernel/git/jejb/scsi.git" and kernel/git/mkp/scsi.git", and I mistakenly thought there were no new changes in latest Linus tree. Thanks, Xingui .
diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c index 05df7ea6954a..8f9e12f460c3 100644 --- a/drivers/ata/libata-eh.c +++ b/drivers/ata/libata-eh.c @@ -149,6 +149,8 @@ ata_eh_cmd_timeout_table[ATA_EH_CMD_TIMEOUT_TABLE_SIZE] = { .timeouts = ata_eh_flush_timeouts }, { .commands = CMDS(ATA_CMD_VERIFY), .timeouts = ata_eh_reset_timeouts }, + { .commands = CMDS(ATA_CMD_STANDBYNOW1), + .timeouts = ata_eh_flush_timeouts }, }; #undef CMDS diff --git a/include/linux/libata.h b/include/linux/libata.h index 96e626d6a7ca..6c0d385b2cde 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -425,7 +425,7 @@ enum { /* This should match the actual table size of * ata_eh_cmd_timeout_table in libata-eh.c. */ - ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 8, + ATA_EH_CMD_TIMEOUT_TABLE_SIZE = 9, /* User visible DMA mask for DMA control. DO NOT renumber. */ ATA_DMA_MASK_ATA = (1 << 0), /* DMA on ATA Disk */
ATA_CMD_STANDBYNOW1 is not in the ata_eh_cmd_timeout_table and falls back to the 5s default (ATA_EH_CMD_DFL_TIMEOUT), which is too short for some drives to complete a STANDBY IMMEDIATE command issued from ata_dev_power_set_standby() via ata_exec_internal(): ata1.00: qc timeout after 5000 msecs (cmd 0xe0) ata1.00: STANDBY IMMEDIATE failed (err_mask=0x4) Reuse the existing ata_eh_flush_timeouts for this command, as a drive entering standby may need to flush its internal cache, making the flush timeout a natural fit. Signed-off-by: Xingui Yang <yangxingui@huawei.com> --- drivers/ata/libata-eh.c | 2 ++ include/linux/libata.h | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-)