On 11/04/2013 03:25 PM, James Bottomley wrote:
> On Mon, 2013-11-04 at 14:36 +0100, Hannes Reinecke wrote:
>> On 10/31/2013 04:49 PM, Christoph Hellwig wrote:
>>> Looks reasonable to me, but a few minor nitpicks:
>>>
>>>> + spin_lock_irqsave(sdev->host->host_lock, flags);
>>>> + if (scsi_host_eh_past_deadline(sdev->host)) {
>>>
>>> I don't have the implementation of scsi_host_eh_past_deadline in my
>>> local tree, but do we really need the host lock for it?
>>>
>> Yes. The eh_deadline variable might be set from an interrupt context
>> or from userland, so we need to protect access to it.
>
> That's not really true. on all our supported architectures 32 bit
> reads/writes are atomic, which means that if one CPU writes a word at
> the same time another reads one, the reader is guaranteed to see either
> the old or the new data. Given the expense of lock cache line bouncing
> on the newer architectures, we really want to avoid a spinlock where
> possible.
>
> In this case, the problem with the implementation is that the writer
> might set eh_deadline to zero, but this is fixable in
> scsi_host_eh_past_deadline() by checking for zero before and after the
> time_before (for the zero to non-zero and non-zero to zero cases).
>
IE you mean something like that attached patch?
Cheers,
Hannes
--
Dr. Hannes Reinecke zSeries & Storage
hare@xxxxxxx +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)
diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index 6a137fa..8abf7ba 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -94,8 +94,10 @@ static int scsi_host_eh_past_deadline(struct Scsi_Host *shost)
if (!shost->last_reset || !shost->eh_deadline)
return 0;
+ /* Double check eh_deadline to catch atomic updates */
if (time_before(jiffies,
- shost->last_reset + shost->eh_deadline))
+ shost->last_reset + shost->eh_deadline) &&
+ shost->eh_deadline)
return 0;
return 1;
@@ -114,15 +116,12 @@ scmd_eh_abort_handler(struct work_struct *work)
unsigned long flags;
int rtn;
- spin_lock_irqsave(sdev->host->host_lock, flags);
if (scsi_host_eh_past_deadline(sdev->host)) {
- spin_unlock_irqrestore(sdev->host->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3,
scmd_printk(KERN_INFO, scmd,
"scmd %p eh timeout, not aborting\n",
scmd));
} else {
- spin_unlock_irqrestore(sdev->host->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3,
scmd_printk(KERN_INFO, scmd,
"aborting command %p\n", scmd));
@@ -1140,16 +1139,13 @@ int scsi_eh_get_sense(struct list_head *work_q,
continue;
shost = scmd->device->host;
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, shost,
"skip %s, past eh deadline\n",
__func__));
break;
}
- spin_unlock_irqrestore(shost->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(2, scmd_printk(KERN_INFO, scmd,
"%s: requesting sense\n",
current->comm));
@@ -1242,19 +1238,15 @@ static int scsi_eh_test_devices(struct list_head *cmd_list,
sdev = scmd->device;
if (!try_stu) {
- spin_lock_irqsave(sdev->host->host_lock, flags);
if (scsi_host_eh_past_deadline(sdev->host)) {
/* Push items back onto work_q */
list_splice_init(cmd_list, work_q);
- spin_unlock_irqrestore(sdev->host->host_lock,
- flags);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, sdev->host,
"skip %s, past eh deadline",
__func__));
break;
}
- spin_unlock_irqrestore(sdev->host->host_lock, flags);
}
finish_cmds = !scsi_device_online(scmd->device) ||
@@ -1301,9 +1293,7 @@ static int scsi_eh_abort_cmds(struct list_head *work_q,
if (!(scmd->eh_eflags & SCSI_EH_CANCEL_CMD))
continue;
shost = scmd->device->host;
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
list_splice_init(&check_list, work_q);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, shost,
@@ -1311,7 +1301,6 @@ static int scsi_eh_abort_cmds(struct list_head *work_q,
__func__));
return list_empty(work_q);
}
- spin_unlock_irqrestore(shost->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3, printk("%s: aborting cmd:"
"0x%p\n", current->comm,
scmd));
@@ -1378,16 +1367,13 @@ static int scsi_eh_stu(struct Scsi_Host *shost,
unsigned long flags;
shost_for_each_device(sdev, shost) {
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, shost,
"skip %s, past eh deadline\n",
__func__));
break;
}
- spin_unlock_irqrestore(shost->host_lock, flags);
stu_scmd = NULL;
list_for_each_entry(scmd, work_q, eh_entry)
if (scmd->device == sdev && SCSI_SENSE_VALID(scmd) &&
@@ -1445,16 +1431,13 @@ static int scsi_eh_bus_device_reset(struct Scsi_Host *shost,
int rtn;
shost_for_each_device(sdev, shost) {
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, shost,
"skip %s, past eh deadline\n",
__func__));
break;
}
- spin_unlock_irqrestore(shost->host_lock, flags);
bdr_scmd = NULL;
list_for_each_entry(scmd, work_q, eh_entry)
if (scmd->device == sdev) {
@@ -1517,9 +1500,7 @@ static int scsi_eh_target_reset(struct Scsi_Host *shost,
unsigned int id;
unsigned long flags;
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
/* push back on work queue for further processing */
list_splice_init(&check_list, work_q);
list_splice_init(&tmp_list, work_q);
@@ -1529,7 +1510,6 @@ static int scsi_eh_target_reset(struct Scsi_Host *shost,
__func__));
return list_empty(work_q);
}
- spin_unlock_irqrestore(shost->host_lock, flags);
scmd = list_entry(tmp_list.next, struct scsi_cmnd, eh_entry);
id = scmd_id(scmd);
@@ -1584,9 +1564,7 @@ static int scsi_eh_bus_reset(struct Scsi_Host *shost,
*/
for (channel = 0; channel <= shost->max_channel; channel++) {
- spin_lock_irqsave(shost->host_lock, flags);
if (scsi_host_eh_past_deadline(shost)) {
- spin_unlock_irqrestore(shost->host_lock, flags);
list_splice_init(&check_list, work_q);
SCSI_LOG_ERROR_RECOVERY(3,
shost_printk(KERN_INFO, shost,
@@ -1594,7 +1572,6 @@ static int scsi_eh_bus_reset(struct Scsi_Host *shost,
__func__));
return list_empty(work_q);
}
- spin_unlock_irqrestore(shost->host_lock, flags);
chan_scmd = NULL;
list_for_each_entry(scmd, work_q, eh_entry) {
