From: Steven Rostedt <rostedt@goodmis.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Ric Wheeler <rwheeler@redhat.com>,
James Bottomley <JBottomley@Parallels.com>
Subject: [12/65] sd: fix array cache flushing bug causing performance problems
Date: Mon, 03 Jun 2013 18:10:28 -0400 [thread overview]
Message-ID: <20130603221154.159834794@goodmis.org> (raw)
In-Reply-To: 20130603221016.243553628@goodmis.org
[-- Attachment #1: 0012-sd-fix-array-cache-flushing-bug-causing-performance-.patch --]
[-- Type: text/plain, Size: 3688 bytes --]
3.6.11.5 stable review patch.
If anyone has any objections, please let me know.
------------------
From: James Bottomley <JBottomley@Parallels.com>
[ Upstream commit 39c60a0948cc06139e2fbfe084f83cb7e7deae3b ]
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>
Cc: <stable@vger.kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
drivers/scsi/sd.c | 20 ++++++++++++++++++++
drivers/scsi/sd.h | 1 +
2 files changed, 21 insertions(+)
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 4df73e5..88ced02 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -141,6 +141,7 @@ 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)
@@ -149,6 +150,13 @@ 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 &&
@@ -161,6 +169,13 @@ 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;
@@ -2127,6 +2142,10 @@ 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)
@@ -2613,6 +2632,7 @@ 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 f703f48..e2b2956 100644
--- a/drivers/scsi/sd.h
+++ b/drivers/scsi/sd.h
@@ -67,6 +67,7 @@ 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 */
--
1.7.10.4
next prev parent reply other threads:[~2013-06-03 22:10 UTC|newest]
Thread overview: 71+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-06-03 22:10 [00/65] 3.6.11.5-stable review Steven Rostedt
2013-06-03 22:10 ` [01/65] KVM: VMX: fix halt emulation while emulating invalid guest sate Steven Rostedt
2013-06-03 22:10 ` [02/65] ARM: OMAP: RX-51: change probe order of touchscreen and panel SPI devices Steven Rostedt
2013-06-03 22:10 ` [03/65] ASoC: wm8994: missing break in wm8994_aif3_hw_params() Steven Rostedt
2013-06-03 22:10 ` [04/65] ACPICA: Fix possible buffer overflow during a field unit read operation Steven Rostedt
2013-06-03 22:10 ` [05/65] ALSA: HDA: Fix Oops caused by dereference NULL pointer Steven Rostedt
2013-06-03 22:10 ` [06/65] dm stripe: fix regression in stripe_width calculation Steven Rostedt
2013-06-03 22:10 ` [07/65] dm bufio: avoid a possible __vmalloc deadlock Steven Rostedt
2013-06-04 12:59 ` Mikulas Patocka
2013-06-04 14:42 ` Steven Rostedt
2013-06-05 14:54 ` Steven Rostedt
2013-06-03 22:10 ` [08/65] dm snapshot: fix error return code in snapshot_ctr Steven Rostedt
2013-06-03 22:10 ` [09/65] math64: New div64_u64_rem helper Steven Rostedt
2013-06-04 13:45 ` Stanislaw Gruszka
2013-06-04 14:43 ` Steven Rostedt
2013-06-03 22:10 ` [10/65] hp_accel: Ignore the error from lis3lv02d_poweron() at resume Steven Rostedt
2013-06-03 22:10 ` [11/65] xen/vcpu/pvhvm: Fix vcpu hotplugging hanging Steven Rostedt
2013-06-03 22:10 ` Steven Rostedt [this message]
2013-06-03 22:10 ` [13/65] audit: Syscall rules are not applied to existing processes on non-x86 Steven Rostedt
2013-06-03 22:10 ` [14/65] timer: Dont reinitialize the cpu base lock during CPU_UP_PREPARE Steven Rostedt
2013-06-03 22:10 ` [15/65] tick: Cleanup NOHZ per cpu data on cpu down Steven Rostedt
2013-06-03 22:10 ` [16/65] tracing: Fix leaks of filter preds Steven Rostedt
2013-06-03 22:10 ` [17/65] ext4: limit group search loop for non-extent files Steven Rostedt
2013-06-03 22:10 ` [18/65] ARM: 7720/1: ARM v6/v7 cmpxchg64 shouldnt clear upper 32 bits of the old/new value Steven Rostedt
2013-06-03 22:10 ` [19/65] powerpc: Bring all threads online prior to migration/hibernation Steven Rostedt
2013-06-03 22:10 ` [20/65] powerpc/kexec: Fix kexec when using VMX optimised memcpy Steven Rostedt
2013-06-03 22:10 ` [21/65] ath9k: fix key allocation error handling for powersave keys Steven Rostedt
2013-06-03 22:10 ` [22/65] mwifiex: clear is_suspended flag when interrupt is received early Steven Rostedt
2013-06-03 22:10 ` [23/65] mwifiex: fix memory leak issue when driver unload Steven Rostedt
2013-06-03 22:10 ` [24/65] mwifiex: fix setting of multicast filter Steven Rostedt
2013-06-03 22:10 ` [25/65] tile: support new Tilera hypervisor Steven Rostedt
2013-06-03 22:10 ` [26/65] B43: Handle DMA RX descriptor underrun Steven Rostedt
2013-06-03 22:10 ` [27/65] drm/mgag200: Fix writes into MGA1064_PIX_CLK_CTL register Steven Rostedt
2013-06-03 22:10 ` [28/65] drm/mgag200: Fix framebuffer base address programming Steven Rostedt
2013-06-03 22:10 ` [29/65] tcp: force a dst refcount when prequeue packet Steven Rostedt
2013-06-03 22:10 ` [30/65] sfc: Fix naming of MTD partitions for FPGA bitfiles Steven Rostedt
2013-06-03 22:10 ` [31/65] net: mac802154: comparision issue of type cast, finding by EXTRA_CFLAGS=-W Steven Rostedt
2013-06-03 22:10 ` [32/65] 3c509.c: call SET_NETDEV_DEV for all device types (ISA/ISAPnP/EISA) Steven Rostedt
2013-06-03 22:10 ` [33/65] net_sched: act_ipt forward compat with xtables Steven Rostedt
2013-06-03 22:10 ` [34/65] net: use netdev_features_t in skb_needs_linearize() Steven Rostedt
2013-06-03 22:10 ` [35/65] net: vlan,ethtool: netdev_features_t is more than 32 bit Steven Rostedt
2013-06-03 22:10 ` [36/65] bridge: fix race with topology change timer Steven Rostedt
2013-06-03 22:10 ` [37/65] packet: tpacket_v3: do not trigger bug() on wrong header status Steven Rostedt
2013-06-03 22:10 ` [38/65] 3c59x: fix freeing nonexistent resource on driver unload Steven Rostedt
2013-06-03 22:10 ` [39/65] 3c59x: fix PCI resource management Steven Rostedt
2013-06-03 22:10 ` [40/65] if_cablemodem.h: Add parenthesis around ioctl macros Steven Rostedt
2013-06-03 22:10 ` [41/65] macvlan: fix passthru mode race between dev removal and rx path Steven Rostedt
2013-06-03 22:10 ` [42/65] ipv6: do not clear pinet6 field Steven Rostedt
2013-06-03 22:10 ` [43/65] xfrm6: release dev before returning error Steven Rostedt
2013-06-03 22:11 ` [44/65] pch_dma: Use GFP_ATOMIC because called from interrupt context Steven Rostedt
2013-06-03 22:11 ` [45/65] watchdog: Fix race condition in registration code Steven Rostedt
2013-06-03 22:11 ` [46/65] drbd: fix for deadlock when using automatic split-brain-recovery Steven Rostedt
2013-06-03 22:11 ` [47/65] ARM: EXYNOS5: Fix kernel dump in AFTR idle mode Steven Rostedt
2013-06-03 22:11 ` [48/65] drivers/rtc/rtc-pcf2123.c: fix error return code in pcf2123_probe() Steven Rostedt
2013-06-03 22:11 ` [49/65] iscsi-target: Fix processing of OOO commands Steven Rostedt
2013-06-03 22:11 ` [50/65] target: close target_put_sess_cmd() vs. core_tmr_abort_task() race Steven Rostedt
2013-06-03 22:11 ` [51/65] ACPI / EC: Restart transaction even when the IBF flag set Steven Rostedt
2013-06-03 22:11 ` [52/65] drivers/char/ipmi: memcpy, need additional 2 bytes to avoid memory overflow Steven Rostedt
2013-06-03 22:11 ` [53/65] ipmi: ipmi_devintf: compat_ioctl method fails to take ipmi_mutex Steven Rostedt
2013-06-03 22:11 ` [54/65] drm/radeon: check incoming cliprects pointer Steven Rostedt
2013-06-03 22:11 ` [55/65] usermodehelper: check subprocess_info->path != NULL Steven Rostedt
2013-06-03 22:11 ` [56/65] parisc: only re-enable interrupts if we need to schedule or deliver signals when returning to userspace Steven Rostedt
2013-06-03 22:11 ` [57/65] parisc: fix SMP races when updating PTE and TLB entries in entry.S Steven Rostedt
2013-06-03 22:11 ` [58/65] parisc: fix NATIVE set up in build Steven Rostedt
2013-06-03 22:11 ` [59/65] parisc: make default cross compiler search more robust (v3) Steven Rostedt
2013-06-03 22:11 ` [60/65] btrfs: dont stop searching after encountering the wrong item Steven Rostedt
2013-06-03 22:11 ` [61/65] hwmon: fix error return code in abituguru_probe() Steven Rostedt
2013-06-03 22:11 ` [62/65] i2c: designware: fix RX FIFO overrun Steven Rostedt
2013-06-03 22:11 ` [63/65] i2c: designware: always clear interrupts before enabling them Steven Rostedt
2013-06-03 22:11 ` [64/65] drm/radeon: Fix VRAM size calculation for VRAM >= 4GB Steven Rostedt
2013-06-03 22:11 ` [65/65] virtio_console: fix uapi header Steven Rostedt
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20130603221154.159834794@goodmis.org \
--to=rostedt@goodmis.org \
--cc=JBottomley@Parallels.com \
--cc=linux-kernel@vger.kernel.org \
--cc=rwheeler@redhat.com \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox