SCSI: sd: fix array cache flushing bug causing performance problems

commit 39c60a0948cc06139e2fbfe084f83cb7e7deae3b upstream.

Some arrays synchronize their full non volatile cache when the sd driver sends
a SYNCHRONIZE CACHE command.  Unfortunately, they can have Terrabytes of this
and we send a SYNCHRONIZE CACHE for every barrier if an array reports it has a
writeback cache.  This leads to massive slowdowns on journalled filesystems.

The fix is to allow userspace to turn off the writeback cache setting as a
temporary measure (i.e. without doing the MODE SELECT to write it back to the
device), so even though the device reported it has a writeback cache, the
user, knowing that the cache is non volatile and all they care about is
filesystem correctness, can turn that bit off in the kernel and avoid the
performance ruinous (and safety irrelevant) SYNCHRONIZE CACHE commands.

The way you do this is add a 'temporary' prefix when performing the usual
cache setting operations, so

echo temporary write through > /sys/class/scsi_disk/<disk>/cache_type

Reported-by: Ric Wheeler <rwheeler@redhat.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 515ec6d64fb43e4a46cfd59a3d3fd1c19cd7c693..7db7eb74dc7806787bff6eb6fd79b14d6ca43b81 100644 (file)
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -138,6 +138,7 @@ sd_store_cache_type(struct device *dev, struct device_attribute *attr,
        char *buffer_data;
        struct scsi_mode_data data;
        struct scsi_sense_hdr sshdr;
+       const char *temp = "temporary ";
        int len;
 
        if (sdp->type != TYPE_DISK)
@@ -146,6 +147,13 @@ sd_store_cache_type(struct device *dev, struct device_attribute *attr,
                 * it's not worth the risk */
                return -EINVAL;
 
+       if (strncmp(buf, temp, sizeof(temp) - 1) == 0) {
+               buf += sizeof(temp) - 1;
+               sdkp->cache_override = 1;
+       } else {
+               sdkp->cache_override = 0;
+       }
+
        for (i = 0; i < ARRAY_SIZE(sd_cache_types); i++) {
                len = strlen(sd_cache_types[i]);
                if (strncmp(sd_cache_types[i], buf, len) == 0 &&
@@ -158,6 +166,13 @@ sd_store_cache_type(struct device *dev, struct device_attribute *attr,
                return -EINVAL;
        rcd = ct & 0x01 ? 1 : 0;
        wce = ct & 0x02 ? 1 : 0;
+
+       if (sdkp->cache_override) {
+               sdkp->WCE = wce;
+               sdkp->RCD = rcd;
+               return count;
+       }
+
        if (scsi_mode_sense(sdp, 0x08, 8, buffer, sizeof(buffer), SD_TIMEOUT,
                            SD_MAX_RETRIES, &data, NULL))
                return -EINVAL;
@@ -2036,6 +2051,10 @@ sd_read_cache_type(struct scsi_disk *sdkp, unsigned char *buffer)
        int old_rcd = sdkp->RCD;
        int old_dpofua = sdkp->DPOFUA;
 
+
+       if (sdkp->cache_override)
+               return;
+
        first_len = 4;
        if (sdp->skip_ms_page_8) {
                if (sdp->type == TYPE_RBC)
@@ -2517,6 +2536,7 @@ static void sd_probe_async(void *data, async_cookie_t cookie)
        sdkp->capacity = 0;
        sdkp->media_present = 1;
        sdkp->write_prot = 0;
+       sdkp->cache_override = 0;
        sdkp->WCE = 0;
        sdkp->RCD = 0;
        sdkp->ATO = 0;

diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h
index 6ad798bfd52a05b0c45de905756e0898c324e3f0..812f1baa54e21f524a4bbb53bf65dfc4e2de2eaa 100644 (file)
--- a/drivers/scsi/sd.h
+++ b/drivers/scsi/sd.h
@@ -70,6 +70,7 @@ struct scsi_disk {
        u8              protection_type;/* Data Integrity Field */
        u8              provisioning_mode;
        unsigned        ATO : 1;        /* state of disk ATO bit */
+       unsigned        cache_override : 1; /* temp override of WCE,RCD */
        unsigned        WCE : 1;        /* state of disk WCE bit */
        unsigned        RCD : 1;        /* state of disk RCD bit, unused */
        unsigned        DPOFUA : 1;     /* state of disk DPOFUA bit */