Hi,
what is the status of write barrier support in Linux?
Is there a web page that documents patches and status?
Thanks in advance,
--
Matthias Andree
On Thu, Jun 27 2002, Matthias Andree wrote:
> Hi,
>
> what is the status of write barrier support in Linux?
We have stable support for IDE (ie block layer barrier support works,
IDE implementation works). I doubt we'll ever do 2.4 SCSI support,
it would be too invasive to really make it safe.
> Is there a web page that documents patches and status?
Not to my knowledge.
--
Jens Axboe
Jens,
I just got crapped all over trying to get us "write barrier" opcodes :-/.
However I do have the start of a draft to submit soon. I could not piggy
back of FUA by MicroSoft last week.
So how did the talk go at OLS for the IDE roadmap to destruction go?
I could not attend, as I was doing other stuff associated with the
industry.
Cheers,
Andre Hedrick
LAD Storage Consulting Group
On Fri, 28 Jun 2002, Jens Axboe wrote:
> On Thu, Jun 27 2002, Matthias Andree wrote:
> > Hi,
> >
> > what is the status of write barrier support in Linux?
>
> We have stable support for IDE (ie block layer barrier support works,
> IDE implementation works). I doubt we'll ever do 2.4 SCSI support,
> it would be too invasive to really make it safe.
>
> > Is there a web page that documents patches and status?
>
> Not to my knowledge.
>
> --
> Jens Axboe
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
On Fri, Jun 28 2002, Andre Hedrick wrote:
>
> Jens,
>
> I just got crapped all over trying to get us "write barrier" opcodes :-/.
> However I do have the start of a draft to submit soon. I could not piggy
> back of FUA by MicroSoft last week.
Basic FUA bit for WRITE command would be good, as long as it also
prevents reordering of the writes currently in write cache. I don't
think mmc makes any such guarentee, although I would have to check to be
sure.
> So how did the talk go at OLS for the IDE roadmap to destruction go?
> I could not attend, as I was doing other stuff associated with the
> industry.
I don't think there was such a talk?! If so, I didn't attend.
--
Jens Axboe
On Fri, Jun 28 2002, Jens Axboe wrote:
> On Fri, Jun 28 2002, Andre Hedrick wrote:
> >
> > Jens,
> >
> > I just got crapped all over trying to get us "write barrier" opcodes :-/.
> > However I do have the start of a draft to submit soon. I could not piggy
> > back of FUA by MicroSoft last week.
>
> Basic FUA bit for WRITE command would be good, as long as it also
> prevents reordering of the writes currently in write cache. I don't
> think mmc makes any such guarentee, although I would have to check to be
> sure.
Oh, and of course options for the tagged commands are needed as well!
--
Jens Axboe
On Fri, 28 Jun 2002, Jens Axboe wrote:
> On Fri, Jun 28 2002, Andre Hedrick wrote:
> >
> > Jens,
> >
> > I just got crapped all over trying to get us "write barrier" opcodes :-/.
> > However I do have the start of a draft to submit soon. I could not piggy
> > back of FUA by MicroSoft last week.
>
> Basic FUA bit for WRITE command would be good, as long as it also
> prevents reordering of the writes currently in write cache. I don't
> think mmc makes any such guarentee, although I would have to check to be
> sure.
Well that is the fuzzy part!
It bypasses the write cache thus "write barrier" is not doable!
In tag, it does not nuke the tags outstanding, it just blows threw the
queue and plunges the meta-data to platter.
Basically worthless unless one is dealing with data-base only.
> > So how did the talk go at OLS for the IDE roadmap to destruction go?
> > I could not attend, as I was doing other stuff associated with the
> > industry.
>
> I don't think there was such a talk?! If so, I didn't attend.
Oh, there was one scheduled but no speakers showed up,
http://www.uzix.org/img_0861.jpg
I was in Irvine during that time for a NCITS meeting thus I cancelled
early thus the name replacement.
Cheers,
Andre Hedrick
LAD Storage Consulting Group
On Fri, 28 Jun 2002, Jens Axboe wrote:
> On Thu, Jun 27 2002, Matthias Andree wrote:
> > Hi,
> >
> > what is the status of write barrier support in Linux?
>
> We have stable support for IDE (ie block layer barrier support works,
> IDE implementation works). I doubt we'll ever do 2.4 SCSI support,
> it would be too invasive to really make it safe.
Alas, my system runs on some nice U160-SCSI drive. Looks like I should
follow 2.5 in the not too distant future then ;-)
OK, thanks.
