| Message ID | 1396399357-28388-1-git-send-email-martin.petersen@oracle.com |
|---|---|
| State | Not Applicable |
| Delegated to: | David Miller |
| Headers | show |
On Tue, Apr 01, 2014 at 08:42:36PM -0400, Martin K. Petersen wrote: > From: "Martin K. Petersen" <martin.petersen@oracle.com> > > Now that drives with support for queued trim are starting to appear, it > would be helpful to expose the chosen trim mode to userland. Create a Can you please be a bit more elaborate with how it'd be helpful? Do we have actual use cases for it? Thanks.
>>>>> "Tejun" == Tejun Heo <tj@kernel.org> writes: >> Now that drives with support for queued trim are starting to appear, >> it would be helpful to expose the chosen trim mode to >> userland. Create a Tejun> Can you please be a bit more elaborate with how it'd be helpful? Tejun> Do we have actual use cases for it? There have been several threads on linux-btrfs and elsewhere where people have asked whether queued trim is supported by their drive or not. This information currently isn't reported anywhere. I thought it would be convenient to have a sysfs attribute that people could inspect to find out.
On Wed, Apr 02, 2014 at 01:26:16PM -0400, Martin K. Petersen wrote: > >>>>> "Tejun" == Tejun Heo <tj@kernel.org> writes: > > >> Now that drives with support for queued trim are starting to appear, > >> it would be helpful to expose the chosen trim mode to > >> userland. Create a > > Tejun> Can you please be a bit more elaborate with how it'd be helpful? > Tejun> Do we have actual use cases for it? > > There have been several threads on linux-btrfs and elsewhere where > people have asked whether queued trim is supported by their drive or > not. This information currently isn't reported anywhere. I thought it > would be convenient to have a sysfs attribute that people could inspect > to find out. Hi Tejun, I haven't had time to pull a new kernel and install the patches I need and this one yet (sorry), but I wanted to give a +1 to that. If queued trim is not available, people may want to disable trim if they want fast performance for rm. But indeed currently it's pretty hard to know if your drive supports it. Thanks, Marc
On Tue, 1 Apr 2014 08:42:36 PM Martin K. Petersen wrote: > Now that drives with support for queued trim are starting to appear, it > would be helpful to expose the chosen trim mode to userland. Create a > sysfs "trim" attribute for each ata_device that displays whether trim is > "unsupported", "unqueued" or "queued". Thank you Martin, this is a really great idea following on from those discussions on the btrfs list on how to work out whether a drive supports trim and, if so, whether it's the original or new variant. I'm hoping to have some time this weekend to test these patches out; should I test them on 3.14 or the current master for 3.15, or both? Much obliged! Chris
On Tue, 1 Apr 2014 08:42:36 PM Martin K. Petersen wrote: > Now that drives with support for queued trim are starting to appear, it > would be helpful to expose the chosen trim mode to userland. Create a > sysfs "trim" attribute for each ata_device that displays whether trim is > "unsupported", "unqueued" or "queued". Both patches applied cleanly to 3.14, kernel now reports "unsupported" for everything other than the SSD in my ZaReason laptop (Crucial M4 mSATA SSD) which reports "unqueued" (which is expected). Tested-By: Chris Samuel <chris@csamuel.org> Thanks! Chris
On Tue, Apr 01, 2014 at 08:42:36PM -0400, Martin K. Petersen wrote: > From: "Martin K. Petersen" <martin.petersen@oracle.com> > > Now that drives with support for queued trim are starting to appear, it > would be helpful to expose the chosen trim mode to userland. Create a > sysfs "trim" attribute for each ata_device that displays whether trim is > "unsupported", "unqueued" or "queued". Ok, I'm going to sound stupid, but which sysfs property is this in? Or more simply what file do I have to cat see this? Thanks, Marc > Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> > Cc: Chris Samuel <chris@csamuel.org> > Cc: Marc MERLIN <marc@merlins.org> > --- > drivers/ata/libata-transport.c | 32 ++++++++++++++++++++++++++++++++ > 1 file changed, 32 insertions(+) > > diff --git a/drivers/ata/libata-transport.c b/drivers/ata/libata-transport.c > index e37413228228..a9a1a9055a3a 100644 > --- a/drivers/ata/libata-transport.c > +++ b/drivers/ata/libata-transport.c > @@ -559,6 +559,37 @@ show_ata_dev_gscr(struct device *dev, > > static DEVICE_ATTR(gscr, S_IRUGO, show_ata_dev_gscr, NULL); > > +enum { > + ATA_TRIM_UNSUPPORTED = 0, > + ATA_TRIM_UNQUEUED = 1, > + ATA_TRIM_QUEUED = 2, > +}; > + > +static const char *trim_mode[] = { > + [ATA_TRIM_UNSUPPORTED] = "unsupported", > + [ATA_TRIM_UNQUEUED] = "unqueued", > + [ATA_TRIM_QUEUED] = "queued", > +}; > + > +static ssize_t > +show_ata_dev_trim(struct device *dev, > + struct device_attribute *attr, char *buf) > +{ > + struct ata_device *ata_dev = transport_class_to_dev(dev); > + unsigned int mode; > + > + if (!ata_id_has_trim(ata_dev->id)) > + mode = ATA_TRIM_UNSUPPORTED; > + else if (ata_fpdma_dsm_supported(ata_dev)) > + mode = ATA_TRIM_QUEUED; > + else > + mode = ATA_TRIM_UNQUEUED; > + > + return snprintf(buf, 20, "%s\n", trim_mode[mode]); > +} > + > +static DEVICE_ATTR(trim, S_IRUGO, show_ata_dev_trim, NULL); > + > static DECLARE_TRANSPORT_CLASS(ata_dev_class, > "ata_device", NULL, NULL, NULL); > > @@ -732,6 +763,7 @@ struct scsi_transport_template *ata_attach_transport(void) > SETUP_DEV_ATTRIBUTE(ering); > SETUP_DEV_ATTRIBUTE(id); > SETUP_DEV_ATTRIBUTE(gscr); > + SETUP_DEV_ATTRIBUTE(trim); > BUG_ON(count > ATA_DEV_ATTRS); > i->dev_attrs[count] = NULL; > > -- > 1.8.3.1 > >
On Sun, Apr 06, 2014 at 03:05:51PM +1000, Chris Samuel wrote: > On Sat, 5 Apr 2014 09:03:12 PM you wrote: > > > Ok, I'm going to sound stupid, but which sysfs property is this in? > > Or more simply what file do I have to cat see this? > > I found them with a simple: > > find /sys -name trim > > I looked at my dmesg info from boot to correlate with devices. Ah, that helps, thanks. So here's another Tested-By: Chris Samuel <marc@merlins.org> legolas [mc]# cat /sys/devices/pci0000:00/0000:00:1f.2/ata1/link1/dev1.0/ata_device/dev1.0/trim unqueued :( So much for the Samsung 840 EVO then. Then again, it does confirm why deleting a kernel tree takes 1.5 seconds. legolas [mc]# smartctl -i /dev/sda smartctl 6.2 2013-07-26 r3841 [x86_64-linux-3.14.0-amd64-i915-preempt-20140216] (local build) Copyright (C) 2002-13, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Device Model: Samsung SSD 840 EVO 1TB Serial Number: S1D9NEAD934600N LU WWN Device Id: 5 002538 85009a8ff Firmware Version: EXT0BB0Q User Capacity: 1,000,204,886,016 bytes [1.00 TB] Sector Size: 512 bytes logical/physical Rotation Rate: Solid State Device Device is: Not in smartctl database [for details use: -P showall] ATA Version is: ACS-2, ATA8-ACS T13/1699-D revision 4c SATA Version is: SATA 3.1, 6.0 Gb/s (current: 6.0 Gb/s) Local Time is: Sat Apr 5 23:18:08 2014 PDT SMART support is: Available - device has SMART capability. SMART support is: Enabled Thanks for the new code. Marc
>>>>> "Chris" == Chris Samuel <chris@csamuel.org> writes:
Chris> Both patches applied cleanly to 3.14, kernel now reports
Chris> "unsupported" for everything other than the SSD in my ZaReason
Chris> laptop (Crucial M4 mSATA SSD) which reports "unqueued" (which is
Chris> expected).
Chris> Tested-By: Chris Samuel <chris@csamuel.org>
Thanks for testing (both of you)!
Hello, On Tue, Apr 01, 2014 at 08:42:36PM -0400, Martin K. Petersen wrote: > +static ssize_t > +show_ata_dev_trim(struct device *dev, > + struct device_attribute *attr, char *buf) > +{ > + struct ata_device *ata_dev = transport_class_to_dev(dev); > + unsigned int mode; Let's please just do const char *mode = "unsupported"; if (ata_id_has_trim(ata_dev->id)) { if (ata_fpdma_dsm_supported(ata_dev)) mode = "queued"; else mode = "unqueued"; } return sprintf(buf, "%s\n", mode); Thanks.
diff --git a/drivers/ata/libata-transport.c b/drivers/ata/libata-transport.c index e37413228228..a9a1a9055a3a 100644 --- a/drivers/ata/libata-transport.c +++ b/drivers/ata/libata-transport.c @@ -559,6 +559,37 @@ show_ata_dev_gscr(struct device *dev, static DEVICE_ATTR(gscr, S_IRUGO, show_ata_dev_gscr, NULL); +enum { + ATA_TRIM_UNSUPPORTED = 0, + ATA_TRIM_UNQUEUED = 1, + ATA_TRIM_QUEUED = 2, +}; + +static const char *trim_mode[] = { + [ATA_TRIM_UNSUPPORTED] = "unsupported", + [ATA_TRIM_UNQUEUED] = "unqueued", + [ATA_TRIM_QUEUED] = "queued", +}; + +static ssize_t +show_ata_dev_trim(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct ata_device *ata_dev = transport_class_to_dev(dev); + unsigned int mode; + + if (!ata_id_has_trim(ata_dev->id)) + mode = ATA_TRIM_UNSUPPORTED; + else if (ata_fpdma_dsm_supported(ata_dev)) + mode = ATA_TRIM_QUEUED; + else + mode = ATA_TRIM_UNQUEUED; + + return snprintf(buf, 20, "%s\n", trim_mode[mode]); +} + +static DEVICE_ATTR(trim, S_IRUGO, show_ata_dev_trim, NULL); + static DECLARE_TRANSPORT_CLASS(ata_dev_class, "ata_device", NULL, NULL, NULL); @@ -732,6 +763,7 @@ struct scsi_transport_template *ata_attach_transport(void) SETUP_DEV_ATTRIBUTE(ering); SETUP_DEV_ATTRIBUTE(id); SETUP_DEV_ATTRIBUTE(gscr); + SETUP_DEV_ATTRIBUTE(trim); BUG_ON(count > ATA_DEV_ATTRS); i->dev_attrs[count] = NULL;