libata: add @ap to ata_wait_register() and introduce ata_msleep()
PATCH 1/3 libata: add @ap to ata_wait_register() and introduce ata_msleep()
Add optional @ap argument to ata_wait_register() and replace msleep() calls with ata_msleep() which take optional @ap in addition to the duration. These will be used to implement EH exclusion.
This patch doesn't cause any behavior difference.
Signed-off-by Tejun Heo <[email protected]>
---
These three patches fix the long-standing EH exclusion bug which was reported more than three years ago and recently caused obscure problems during parallel probing for ata_piix w/ SIDPR.
It basically implements EH big lock which guarantees that only single EH of a host is running. This achieves proper EH exclusion without pervasive changes to each driver which would be fragile while not harming parallel probing/EH too much.
Thanks.
drivers/ata/libahci.c | 18 +++++++++--------- drivers/ata/libata-core.c | 21 ++++++++++++++------- drivers/ata/libata-eh.c | 2 +- drivers/ata/libata-sff.c | 12 ++++++------ drivers/ata/pata_bf54x.c | 4 ++-- drivers/ata/pata_samsung_cf.c | 2 +- drivers/ata/pata_scc.c | 4 ++-- drivers/ata/sata_fsl.c | 19 ++++++++++--------- drivers/ata/sata_inic162x.c | 2 +- drivers/ata/sata_sil24.c | 14 +++++++------- drivers/ata/sata_via.c | 2 +- include/linux/libata.h | 5 +++-- 12 files changed, 57 insertions(+), 48 deletions(-)
drivers/ata/libahci.c +9 -9
@@ -577,7 +577,7 @@ int ahci_stop_engine(struct ata_port *ap)writel (tmp, port_mmio +PORT_CMD );/* wait for engine to stop. This could be as long as 500 msec */ - tmp =ata_wait_register (port_mmio +PORT_CMD , + tmp =ata_wait_register (ap, port_mmio +PORT_CMD ,PORT_CMD_LIST_ON ,PORT_CMD_LIST_ON ,1 ,500 );if (tmp &PORT_CMD_LIST_ON )return -EIO ;
@@ -624,7 +624,7 @@ static int ahci_stop_fis_rx(struct ata_port *ap)writel (tmp, port_mmio +PORT_CMD );/* wait for completion, spec says 500ms, give it 1000 */ - tmp =ata_wait_register (port_mmio +PORT_CMD ,PORT_CMD_FIS_ON , + tmp =ata_wait_register (ap, port_mmio +PORT_CMD ,PORT_CMD_FIS_ON ,PORT_CMD_FIS_ON ,10 ,1000 );if (tmp &PORT_CMD_FIS_ON )return -EBUSY ;
@@ -673,7 +673,7 @@ static void ahci_disable_alpm(struct ata_port *ap) cmd =readl (port_mmio +PORT_CMD );/* wait 10ms to be sure we've come out of any low power state */ -msleep (10 ); +ata_msleep (ap,10 );/* clear out any PhyRdy stuff from interrupt status */ writel (PORT_IRQ_PHYRDY , port_mmio +PORT_IRQ_STAT );
@@ -821,7 +821,7 @@ static void ahci_start_port(struct ata_port *ap) emp->led_state, 4);if (rc == -EBUSY ) -msleep (1 ); +ata_msleep (ap,1 );else break ; }
@@ -880,7 +880,7 @@ int ahci_reset_controller(struct ata_host *host)* reset must complete within 1 second, or * the hardware should be considered fried. */ - tmp =ata_wait_register (mmio +HOST_CTL ,HOST_RESET , + tmp =ata_wait_register (NULL , mmio +HOST_CTL ,HOST_RESET ,HOST_RESET ,10 ,1000 );if (tmp &HOST_RESET ) {
@@ -1260,7 +1260,7 @@ int ahci_kick_engine(struct ata_port *ap)writel (tmp, port_mmio +PORT_CMD ); rc =0 ; - tmp =ata_wait_register (port_mmio +PORT_CMD , + tmp =ata_wait_register (ap, port_mmio +PORT_CMD ,PORT_CMD_CLO ,PORT_CMD_CLO ,1 ,500 );if (tmp &PORT_CMD_CLO ) rc = -EIO ;
@@ -1290,8 +1290,8 @@ static int ahci_exec_polled_cmd(struct ata_port *ap, int pmp,writel (1 , port_mmio +PORT_CMD_ISSUE );if (timeout_msec) { - tmp =ata_wait_register (port_mmio +PORT_CMD_ISSUE ,0x1 ,0x1 , -1 , timeout_msec); + tmp =ata_wait_register (ap, port_mmio +PORT_CMD_ISSUE , +0x1 ,0x1 ,1 , timeout_msec);if (tmp &0x1 ) {ahci_kick_engine (ap);return -EBUSY ;
@@ -1338,7 +1338,7 @@ int ahci_do_softreset(struct ata_link *link, unsigned int *class, }/* spec says at least 5us, but be generous and sleep for 1ms */ -msleep (1 ); +ata_msleep (ap,1 );/* issue the second D2H Register FIS */ tf.ctl &= ~ATA_SRST ;
drivers/ata/libata-core.c +14 -7
@@ -3596,7 +3596,7 @@ int ata_wait_ready(struct ata_link *link, unsigned long deadline, warned =1 ; } -msleep (50 ); +ata_msleep (link->ap,50 ); } }
@@ -3617,7 +3617,7 @@ int ata_wait_ready(struct ata_link *link, unsigned long deadline,int ata_wait_after_reset (struct ata_link *link,unsigned long deadline,int (*check_ready)(struct ata_link *link)) { -msleep (ATA_WAIT_AFTER_RESET ); +ata_msleep (link->ap,ATA_WAIT_AFTER_RESET );return ata_wait_ready (link, deadline, check_ready); }
@@ -3665,7 +3665,7 @@ int sata_link_debounce(struct ata_link *link, const unsigned long *params, last_jiffies = jiffies;while (1 ) { -msleep (interval); +ata_msleep (link->ap, interval);if ((rc =sata_scr_read (link,SCR_STATUS , &cur)))return rc; cur &=0xf ;
@@ -3730,7 +3730,7 @@ int sata_link_resume(struct ata_link *link, const unsigned long *params,* immediately after resuming. Delay 200ms before * debouncing. */ -msleep (200 ); +ata_msleep (link->ap,200 );/* is SControl restored correctly? */ if ((rc =sata_scr_read (link,SCR_CONTROL , &scontrol)))
@@ -3868,7 +3868,7 @@ int sata_link_hardreset(struct ata_link *link, const unsigned long *timing,/* Couldn't find anything in SATA I/II specs, but AHCI-1.1 * 10.4.2 says at least 1 ms. */ -msleep (1 ); +ata_msleep (link->ap,1 );/* bring link back */ rc =sata_link_resume (link, timing, deadline);
@@ -6606,8 +6606,14 @@ int ata_ratelimit(void)return __ratelimit (&ratelimit); } +void ata_msleep (struct ata_port *ap,unsigned int msecs) +{ +msleep (msecs); +} +/** * ata_wait_register - wait until register value changes +* @ap: ATA port to wait register for, can be NULL * @reg: IO-mapped register * @mask: Mask to apply to read register value * @val: Wait condition
@@ -6629,7 +6635,7 @@ int ata_ratelimit(void)* RETURNS: * The final register value. */ -u32 ata_wait_register (void __iomem *reg,u32 mask,u32 val, +u32 ata_wait_register (struct ata_port *ap,void __iomem *reg,u32 mask,u32 val,unsigned long interval,unsigned long timeout) {unsigned long deadline;
@@ -6644,7 +6650,7 @@ u32 ata_wait_register(void __iomem *reg, u32 mask, u32 val, deadline =ata_deadline (jiffies, timeout);while ((tmp & mask) == val &&time_before (jiffies, deadline)) { -msleep (interval); +ata_msleep (ap, interval); tmp =ioread32 (reg); }
@@ -6727,6 +6733,7 @@ EXPORT_SYMBOL_GPL(ata_std_postreset);EXPORT_SYMBOL_GPL (ata_dev_classify);EXPORT_SYMBOL_GPL (ata_dev_pair);EXPORT_SYMBOL_GPL (ata_ratelimit); +EXPORT_SYMBOL_GPL (ata_msleep);EXPORT_SYMBOL_GPL (ata_wait_register);EXPORT_SYMBOL_GPL (ata_scsi_queuecmd);EXPORT_SYMBOL_GPL (ata_scsi_slave_config);
drivers/ata/libata-eh.c +1 -1
@@ -779,7 +779,7 @@ void ata_port_wait_eh(struct ata_port *ap)/* make sure SCSI EH is complete */ if (scsi_host_in_recovery (ap->scsi_host)) { -msleep (10 ); +ata_msleep (ap,10 ); goto retry; } }
drivers/ata/libata-sff.c +6 -6
@@ -222,7 +222,7 @@ int ata_sff_busy_sleep(struct ata_port *ap, timeout =ata_deadline (timer_start, tmout_pat);while (status !=0xff && (status &ATA_BUSY ) &&time_before (jiffies, timeout)) { -msleep (50 ); +ata_msleep (ap,50 ); status =ata_sff_busy_wait (ap,ATA_BUSY ,3 ); }
@@ -234,7 +234,7 @@ int ata_sff_busy_sleep(struct ata_port *ap, timeout =ata_deadline (timer_start, tmout);while (status !=0xff && (status &ATA_BUSY ) &&time_before (jiffies, timeout)) { -msleep (50 ); +ata_msleep (ap,50 ); status = ap->ops->sff_check_status (ap); }
@@ -360,7 +360,7 @@ static void ata_dev_select(struct ata_port *ap, unsigned int device,if (wait) {if (can_sleep && ap->link.device[device].class ==ATA_DEV_ATAPI ) -msleep (150 ); +ata_msleep (ap,150 );ata_wait_idle (ap); } }
@@ -1342,7 +1342,7 @@ fsm_start:*/ status =ata_sff_busy_wait (ap,ATA_BUSY ,5 );if (status &ATA_BUSY ) { -msleep (2 ); +ata_msleep (ap,2 ); status =ata_sff_busy_wait (ap,ATA_BUSY ,10 );if (status &ATA_BUSY ) {ata_sff_queue_pio_task (ap,ATA_SHORT_PAUSE );
@@ -1917,7 +1917,7 @@ int ata_sff_wait_after_reset(struct ata_link *link, unsigned int devmask,unsigned int dev1 = devmask & (1 <<1 );int rc, ret =0 ; -msleep (ATA_WAIT_AFTER_RESET ); +ata_msleep (ap,ATA_WAIT_AFTER_RESET );/* always check readiness of the master device */ rc =ata_sff_wait_ready (link, deadline);
@@ -1946,7 +1946,7 @@ int ata_sff_wait_after_reset(struct ata_link *link, unsigned int devmask, lbal =ioread8 (ioaddr->lbal_addr);if ((nsect ==1 ) && (lbal ==1 ))break ; -msleep (50 );/* give drive a breather */ +ata_msleep (ap,50 );/* give drive a breather */ } rc =ata_sff_wait_ready (link, deadline);
drivers/ata/pata_bf54x.c +2 -2
@@ -1046,7 +1046,7 @@ static void bfin_bus_post_reset(struct ata_port *ap, unsigned int devmask) dev1 =0 ;break ; } -msleep (50 );/* give drive a breather */ +ata_msleep (ap,50 );/* give drive a breather */ }if (dev1)ata_sff_busy_sleep (ap,ATA_TMOUT_BOOT_QUICK ,ATA_TMOUT_BOOT );
@@ -1087,7 +1087,7 @@ static unsigned int bfin_bus_softreset(struct ata_port *ap,* * Old drivers/ide uses the 2mS rule and then waits for ready */ -msleep (150 ); +ata_msleep (ap,150 );/* Before we perform post reset processing we want to see if * the bus shows 0xFF because the odd clown forgets the D7
drivers/ata/pata_samsung_cf.c +1 -1
@@ -322,7 +322,7 @@ static int pata_s3c_wait_after_reset(struct ata_link *link, {int rc; -msleep (ATA_WAIT_AFTER_RESET ); +ata_msleep (link->ap,ATA_WAIT_AFTER_RESET );/* always check readiness of the master device */ rc =ata_sff_wait_ready (link, deadline);
drivers/ata/pata_scc.c +2 -2
@@ -530,7 +530,7 @@ static int scc_wait_after_reset(struct ata_link *link, unsigned int devmask,* * Old drivers/ide uses the 2mS rule and then waits for ready. */ -msleep (150 ); +ata_msleep (ap,150 );/* always check readiness of the master device */ rc =ata_sff_wait_ready (link, deadline);
@@ -559,7 +559,7 @@ static int scc_wait_after_reset(struct ata_link *link, unsigned int devmask, lbal =in_be32 (ioaddr->lbal_addr);if ((nsect ==1 ) && (lbal ==1 ))break ; -msleep (50 );/* give drive a breather */ +ata_msleep (ap,50 );/* give drive a breather */ } rc =ata_sff_wait_ready (link, deadline);
drivers/ata/sata_fsl.c +10 -9
@@ -678,7 +678,7 @@ static void sata_fsl_port_stop(struct ata_port *ap)iowrite32 (temp, hcr_base +HCONTROL );/* Poll for controller to go offline - should happen immediately */ -ata_wait_register (hcr_base +HSTATUS ,ONLINE ,ONLINE ,1 ,1 ); +ata_wait_register (ap, hcr_base +HSTATUS ,ONLINE ,ONLINE ,1 ,1 ); ap->private_data =NULL ; dma_free_coherent(dev, SATA_FSL_PORT_PRIV_DMA_SZ,
@@ -729,7 +729,8 @@ try_offline_again:iowrite32 (temp, hcr_base +HCONTROL );/* Poll for controller to go offline */ - temp =ata_wait_register (hcr_base +HSTATUS ,ONLINE ,ONLINE ,1 ,500 ); + temp =ata_wait_register (ap, hcr_base +HSTATUS ,ONLINE ,ONLINE , +1 ,500 );if (temp &ONLINE ) { ata_port_printk(ap, KERN_ERR,
@@ -752,7 +753,7 @@ try_offline_again:/* * PHY reset should remain asserted for atleast 1ms */ -msleep (1 ); +ata_msleep (ap,1 );/* * Now, bring the host controller online again, this can take time
@@ -766,7 +767,7 @@ try_offline_again: temp |=HCONTROL_PMP_ATTACHED ;iowrite32 (temp, hcr_base +HCONTROL ); - temp =ata_wait_register (hcr_base +HSTATUS ,ONLINE ,0 ,1 ,500 ); + temp =ata_wait_register (ap, hcr_base +HSTATUS ,ONLINE ,0 ,1 ,500 );if (!(temp &ONLINE )) { ata_port_printk(ap, KERN_ERR,
@@ -784,7 +785,7 @@ try_offline_again:* presence */ - temp =ata_wait_register (hcr_base +HSTATUS ,0xFF ,0 ,1 ,500 ); + temp =ata_wait_register (ap, hcr_base +HSTATUS ,0xFF ,0 ,1 ,500 );if ((!(temp &0x10 )) ||ata_link_offline (link)) { ata_port_printk(ap, KERN_WARNING, "No Device OR PHYRDY change,Hstatus = 0x%x\n",
@@ -797,7 +798,7 @@ try_offline_again:* Wait for the first D2H from device,i.e,signature update notification */ start_jiffies = jiffies; - temp =ata_wait_register (hcr_base +HSTATUS ,0xFF ,0x10 , + temp =ata_wait_register (ap, hcr_base +HSTATUS ,0xFF ,0x10 ,500 ,jiffies_to_msecs (deadline - start_jiffies));if ((temp &0xFF ) !=0x18 ) {
@@ -880,7 +881,7 @@ static int sata_fsl_softreset(struct ata_link *link, unsigned int *class,iowrite32 (pmp,CQPMP + hcr_base);iowrite32 (1 ,CQ + hcr_base); - temp =ata_wait_register (CQ + hcr_base,0x1 ,0x1 ,1 ,5000 ); + temp =ata_wait_register (ap,CQ + hcr_base,0x1 ,0x1 ,1 ,5000 );if (temp &0x1 ) {ata_port_printk (ap,KERN_WARNING ,"ATA_SRST issue failed\n" );
@@ -896,7 +897,7 @@ static int sata_fsl_softreset(struct ata_link *link, unsigned int *class, goto err; } -msleep (1 ); +ata_msleep (ap,1 );/* * SATA device enters reset state after receving a Control register
@@ -915,7 +916,7 @@ static int sata_fsl_softreset(struct ata_link *link, unsigned int *class,if (pmp !=SATA_PMP_CTRL_PORT )iowrite32 (pmp,CQPMP + hcr_base);iowrite32 (1 ,CQ + hcr_base); -msleep (150 );/* ?? */ +ata_msleep (ap,150 );/* ?? */ /* * The above command would have signalled an interrupt on command
drivers/ata/sata_inic162x.c +1 -1
@@ -614,7 +614,7 @@ static int inic_hardreset(struct ata_link *link, unsigned int *class,writew (IDMA_CTL_RST_ATA , idma_ctl);readw (idma_ctl);/* flush */ -msleep (1 ); +ata_msleep (ap,1 );writew (0 , idma_ctl); rc =sata_link_resume (link, timing, deadline);
drivers/ata/sata_sil24.c +7 -7
@@ -589,9 +589,9 @@ static int sil24_init_port(struct ata_port *ap)sil24_clear_pmp (ap);writel (PORT_CS_INIT , port +PORT_CTRL_STAT ); -ata_wait_register (port +PORT_CTRL_STAT , +ata_wait_register (ap, port +PORT_CTRL_STAT ,PORT_CS_INIT ,PORT_CS_INIT ,10 ,100 ); - tmp =ata_wait_register (port +PORT_CTRL_STAT , + tmp =ata_wait_register (ap, port +PORT_CTRL_STAT ,PORT_CS_RDY ,0 ,10 ,100 );if ((tmp & (PORT_CS_INIT |PORT_CS_RDY )) !=PORT_CS_RDY ) {
@@ -631,7 +631,7 @@ static int sil24_exec_polled_cmd(struct ata_port *ap, int pmp,writel ((u64 )paddr >>32 , port +PORT_CMD_ACTIVATE +4 ); irq_mask = (PORT_IRQ_COMPLETE |PORT_IRQ_ERROR ) <<PORT_IRQ_RAW_SHIFT ; - irq_stat =ata_wait_register (port +PORT_IRQ_STAT , irq_mask,0x0 , + irq_stat =ata_wait_register (ap, port +PORT_IRQ_STAT , irq_mask,0x0 ,10 , timeout_msec);writel (irq_mask, port +PORT_IRQ_STAT );/* clear IRQs */
@@ -719,9 +719,9 @@ static int sil24_hardreset(struct ata_link *link, unsigned int *class, "state, performing PORT_RST\n");writel (PORT_CS_PORT_RST , port +PORT_CTRL_STAT ); -msleep (10 ); +ata_msleep (ap,10 );writel (PORT_CS_PORT_RST , port +PORT_CTRL_CLR ); -ata_wait_register (port +PORT_CTRL_STAT ,PORT_CS_RDY ,0 , +ata_wait_register (ap, port +PORT_CTRL_STAT ,PORT_CS_RDY ,0 ,10 ,5000 );/* restore port configuration */
@@ -740,7 +740,7 @@ static int sil24_hardreset(struct ata_link *link, unsigned int *class, tout_msec =5000 ;writel (PORT_CS_DEV_RST , port +PORT_CTRL_STAT ); - tmp =ata_wait_register (port +PORT_CTRL_STAT , + tmp =ata_wait_register (ap, port +PORT_CTRL_STAT ,PORT_CS_DEV_RST ,PORT_CS_DEV_RST ,10 , tout_msec);
@@ -1253,7 +1253,7 @@ static void sil24_init_controller(struct ata_host *host) tmp =readl (port +PORT_CTRL_STAT );if (tmp &PORT_CS_PORT_RST ) {writel (PORT_CS_PORT_RST , port +PORT_CTRL_CLR ); - tmp =ata_wait_register (port +PORT_CTRL_STAT , + tmp =ata_wait_register (NULL , port +PORT_CTRL_STAT ,PORT_CS_PORT_RST ,PORT_CS_PORT_RST ,10 ,100 );if (tmp &PORT_CS_PORT_RST )
drivers/ata/sata_via.c +1 -1
@@ -349,7 +349,7 @@ static int vt6420_prereset(struct ata_link *link, unsigned long deadline)/* wait for phy to become ready, if necessary */ do { - msleep(200); + ata_msleep(link->ap, 200); svia_scr_read(link, SCR_STATUS, &sstatus); if ((sstatus & 0xf) != 1) break;
include/linux/libata.h +3 -2
@@ -994,8 +994,9 @@ extern int ata_host_suspend(struct ata_host *host, pm_message_t mesg);extern void ata_host_resume (struct ata_host *host);#endif extern int ata_ratelimit (void ); -extern u32 ata_wait_register (void __iomem *reg,u32 mask,u32 val, -unsigned long interval,unsigned long timeout); +extern void ata_msleep (struct ata_port *ap,unsigned int msecs); +extern u32 ata_wait_register (struct ata_port *ap,void __iomem *reg,u32 mask, +u32 val,unsigned long interval,unsigned long timeout);extern int atapi_cmd_type (u8 opcode);extern void ata_tf_to_fis (const struct ata_taskfile *tf,u8 pmp,int is_cmd,u8 *fis);
Signature · 1 line
1.7.1
PATCH 2/3 libata: implement cross-port EH exclusion
In libata, the non-EH code paths should always take and release ap->lock explicitly when accessing hardware or shared data structures. However, once EH is active, it's assumed that the port is owned by EH and EH methods don't explicitly take ap->lock unless race from irq handler or other code paths are expected. However, libata EH didn't guarantee exclusion among EHs for ports of the same host. IOW, multiple EHs may execute in parallel on multiple ports of the same controller.
In many cases, especially in SATA, the ports are completely independent of each other and this doesn't cause problems; however, there are cases where different ports share the same resource, which lead to obscure timing related bugs such as the one fixed by commit 213373cf (ata_piix: fix locking around SIDPR access).
This patch implements exclusion among EHs of the same host. When EH begins, it acquires per-host EH ownership by calling ata_eh_acquire(). When EH finishes, the ownership is released by calling ata_eh_release(). EH ownership is also released whenever the EH thread goes to sleep from ata_msleep() or explicitly and reacquired after waking up.
This ensures that while EH is actively accessing the hardware, it has exclusive access to it while allowing EHs to interleave and progress in parallel as they hit waiting stages, which dominate the time spent in EH. This achieves cross-port EH exclusion without pervasive and fragile changes while still allowing parallel EH for the most part.
This was first reported by [email protected] more than three years ago in the following bugzilla. :-)
https://bugzilla.kernel.org/show_bug.cgi?id=8223
Signed-off-by Tejun Heo <[email protected]>
Cc Alan Cox <[email protected]>
Reported-by [email protected]
--- drivers/ata/libata-core.c | 30 ++++++++++++++++++++++++++++++ drivers/ata/libata-eh.c | 44 +++++++++++++++++++++++++++++++++++++++++++- drivers/ata/libata.h | 2 ++ include/linux/libata.h | 5 +++++ 4 files changed, 80 insertions(+), 1 deletions(-)
drivers/ata/libata-core.c +30 -0
@@ -1806,8 +1806,14 @@ unsigned ata_exec_internal_sg(struct ata_device *dev, } } +if (ap->ops->error_handler) +ata_eh_release (ap); + rc =wait_for_completion_timeout (&wait,msecs_to_jiffies (timeout)); +if (ap->ops->error_handler) +ata_eh_acquire (ap); +ata_sff_flush_pio_task (ap);if (!rc) {
@@ -5693,6 +5699,7 @@ struct ata_host *ata_host_alloc(struct device *dev, int max_ports)dev_set_drvdata (dev, host);spin_lock_init (&host->lock); +mutex_init (&host->eh_mutex); host->dev = dev; host->n_ports = max_ports;
@@ -5990,6 +5997,7 @@ void ata_host_init(struct ata_host *host, struct device *dev, unsigned long flags, struct ata_port_operations *ops) { spin_lock_init(&host->lock); +mutex_init (&host->eh_mutex); host->dev = dev; host->flags = flags; host->ops = ops;
@@ -6606,9 +6614,31 @@ int ata_ratelimit(void)return __ratelimit (&ratelimit); } +/** +* ata_msleep - ATA EH owner aware msleep +* @ap: ATA port to attribute the sleep to +* @msecs: duration to sleep in milliseconds +* +* Sleeps @msecs. If the current task is owner of @ap's EH, the +* ownership is released before going to sleep and reacquired +* after the sleep is complete. IOW, other ports sharing the +* @ap->host will be allowed to own the EH while this task is +* sleeping. +* +* LOCKING: +* Might sleep. +*/ void ata_msleep (struct ata_port *ap,unsigned int msecs) { +bool owns_eh = ap && ap->host->eh_owner == current; + +if (owns_eh) +ata_eh_release (ap); +msleep (msecs); + +if (owns_eh) +ata_eh_acquire (ap); }/**
drivers/ata/libata-eh.c +43 -1
82 lines not on this page · raw message
drivers/ata/libata.h +2 -0
@@ -145,6 +145,8 @@ extern int ata_scsi_user_scan(struct Scsi_Host *shost, unsigned int channel,/* libata-eh.c */ extern unsigned long ata_internal_cmd_timeout (struct ata_device *dev,u8 cmd);extern void ata_internal_cmd_timed_out (struct ata_device *dev,u8 cmd); +extern void ata_eh_acquire (struct ata_port *ap); +extern void ata_eh_release (struct ata_port *ap);extern enum blk_eh_timer_return ata_scsi_timed_out (struct scsi_cmnd *cmd);extern void ata_scsi_error (struct Scsi_Host *host);extern void ata_port_wait_eh (struct ata_port *ap);
include/linux/libata.h +5 -0
@@ -37,6 +37,7 @@#include <scsi/scsi_host.h> #include <linux/acpi.h> #include <linux/cdrom.h> +#include <linux/sched.h> /* * Define if arch has non-standard setup. This is a _PCI_ standard
@@ -529,6 +530,10 @@ struct ata_host {void *private_data;struct ata_port_operations *ops;unsigned long flags; + +struct mutex eh_mutex; +struct task_struct *eh_owner; +#ifdef CONFIG_ATA_ACPI acpi_handle acpi_handle;#endif
Signature · 1 line
1.7.1
PATCH 3/3 ata_piix: remove SIDPR locking
Now that libata provides proper cross-port EH exclusion. The SIDPR locking added by commit 213373cf (ata_piix: fix locking around SIDPR access) is no longer necessary. Remove it.
Signed-off-by Tejun Heo <[email protected]>
--- drivers/ata/ata_piix.c | 8 -------- 1 files changed, 0 insertions(+), 8 deletions(-)
drivers/ata/ata_piix.c +0 -8
@@ -158,7 +158,6 @@ struct piix_map_db {struct piix_host_priv {const int *map;u32 saved_iocfg; -spinlock_t sidpr_lock;/* FIXME: remove once locking in EH is fixed */ void __iomem *sidpr; };
@@ -952,15 +951,12 @@ static int piix_sidpr_scr_read(struct ata_link *link, unsigned int reg, u32 *val) { struct piix_host_priv *hpriv = link->ap->host->private_data; -unsigned long flags;if (reg >=ARRAY_SIZE (piix_sidx_map))return -EINVAL ; -spin_lock_irqsave (&hpriv->sidpr_lock, flags);piix_sidpr_sel (link, reg); *val =ioread32 (hpriv->sidpr +PIIX_SIDPR_DATA ); -spin_unlock_irqrestore (&hpriv->sidpr_lock, flags);return 0 ; }
@@ -968,15 +964,12 @@ static int piix_sidpr_scr_write(struct ata_link *link, unsigned int reg, u32 val) {struct piix_host_priv *hpriv = link->ap->host->private_data; -unsigned long flags;if (reg >=ARRAY_SIZE (piix_sidx_map))return -EINVAL ; -spin_lock_irqsave (&hpriv->sidpr_lock, flags);piix_sidpr_sel (link, reg);iowrite32 (val, hpriv->sidpr +PIIX_SIDPR_DATA ); -spin_unlock_irqrestore (&hpriv->sidpr_lock, flags);return 0 ; }
@@ -1573,7 +1566,6 @@ static int __devinit piix_init_one(struct pci_dev *pdev, hpriv =devm_kzalloc (dev,sizeof (*hpriv),GFP_KERNEL );if (!hpriv)return -ENOMEM ; -spin_lock_init (&hpriv->sidpr_lock);/* Save IOCFG, this will be used for cable detection, quirk * detection and restoration on detach. This is necessary
Signature · 1 line
1.7.1
Jeff Garzik
AppliedQuoted 19 lines · Tejun Heo
Add optional @ap argument to
ata_wait_register()and replacemsleep()calls withata_msleep()which take optional @ap in addition to the duration. These will be used to implement EH exclusion.This patch doesn't cause any behavior difference.
Signed-off-by Tejun Heo<[email protected]>---
These three patches fix the long-standing EH exclusion bug which was reported more than three years ago and recently caused obscure problems during parallel probing forata_piixw/ SIDPR.It basically implements EH big lock which guarantees that only single EH of a host is running. This achieves proper EH exclusion without pervasive changes to each driver which would be fragile while not harming parallel probing/EH too much.
Thanks.
Quoted 13 lines
drivers/ata/libahci.c | 18 +++++++++--------- drivers/ata/libata-core.c | 21 ++++++++++++++------- drivers/ata/libata-eh.c | 2 +- drivers/ata/libata-sff.c | 12 ++++++------ drivers/ata/pata_bf54x.c | 4 ++-- drivers/ata/pata_samsung_cf.c | 2 +- drivers/ata/pata_scc.c | 4 ++-- drivers/ata/sata_fsl.c | 19 ++++++++++--------- drivers/ata/sata_inic162x.c | 2 +- drivers/ata/sata_sil24.c | 14 +++++++------- drivers/ata/sata_via.c | 2 +- include/linux/libata.h | 5 +++-- 12 files changed, 57 insertions(+), 48 deletions(-)
applied