xfs
[Top] [All Lists]

Re: [PATCH v2 5/5] dax: handle media errors in dax_do_io

To: Dan Williams <dan.j.williams@xxxxxxxxx>
Subject: Re: [PATCH v2 5/5] dax: handle media errors in dax_do_io
From: Jeff Moyer <jmoyer@xxxxxxxxxx>
Date: Fri, 15 Apr 2016 15:13:26 -0400
Cc: "Verma\, Vishal L" <vishal.l.verma@xxxxxxxxx>, "hch\@infradead.org" <hch@xxxxxxxxxxxxx>, "jack\@suse.cz" <jack@xxxxxxx>, "axboe\@fb.com" <axboe@xxxxxx>, "linux-nvdimm\@ml01.01.org" <linux-nvdimm@xxxxxxxxxxx>, "david\@fromorbit.com" <david@xxxxxxxxxxxxx>, "linux-kernel\@vger.kernel.org" <linux-kernel@xxxxxxxxxxxxxxx>, "xfs\@oss.sgi.com" <xfs@xxxxxxxxxxx>, "linux-block\@vger.kernel.org" <linux-block@xxxxxxxxxxxxxxx>, "linux-mm\@kvack.org" <linux-mm@xxxxxxxxx>, "viro\@zeniv.linux.org.uk" <viro@xxxxxxxxxxxxxxxxxx>, "linux-fsdevel\@vger.kernel.org" <linux-fsdevel@xxxxxxxxxxxxxxx>, "akpm\@linux-foundation.org" <akpm@xxxxxxxxxxxxxxxxxxxx>, "linux-ext4\@vger.kernel.org" <linux-ext4@xxxxxxxxxxxxxxx>, "Wilcox\, Matthew R" <matthew.r.wilcox@xxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <CAPcyv4jcPDDcru1ySJLY7SzDQQWFXbfHm493N0twa1vHBEc6aQ@xxxxxxxxxxxxxx> (Dan Williams's message of "Fri, 15 Apr 2016 11:56:20 -0700")
References: <1459303190-20072-1-git-send-email-vishal.l.verma@xxxxxxxxx> <1459303190-20072-6-git-send-email-vishal.l.verma@xxxxxxxxx> <x49twj26edj.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <1460739288.3012.3.camel@xxxxxxxxx> <x49potq6bm2.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <1460741821.3012.11.camel@xxxxxxxxx> <CAPcyv4hemNM4uQYCPBXyH+DWTOLvyBNBeMYstKbPdad_Cw48HQ@xxxxxxxxxxxxxx> <x49lh4e6928.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <CAPcyv4hRQj2ZsFj7Xa_=OwcHrzP9_5yUpt3LQ+bPH4PcLe7UCQ@xxxxxxxxxxxxxx> <x494mb2ivcl.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <CAPcyv4jcPDDcru1ySJLY7SzDQQWFXbfHm493N0twa1vHBEc6aQ@xxxxxxxxxxxxxx>
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.3 (gnu/linux)
Dan Williams <dan.j.williams@xxxxxxxxx> writes:

> On Fri, Apr 15, 2016 at 11:24 AM, Jeff Moyer <jmoyer@xxxxxxxxxx> wrote:
>>> Moreover, we're going to do the full badblocks lookup anyway when we
>>> call ->direct_access().  If we had that information earlier we can
>>> avoid this fallback dance.
>>
>> None of the proposed approaches looks clean to me.  I'll go along with
>> whatever you guys think is best.  I am in favor of wrapping up all that
>> duplicated code, though.
>
> Christoph originally pushed for open coding this fallback decision
> per-filesystem.  I agree with you on the "none the above" options are
> clean.

I don't recall him saying "open code".  Rather, the sentiment was to
leave the fallback to the callers.  That doesn't mean you can't wrap it
up in a convenience function.

Cheers,
Jeff

<Prev in Thread] Current Thread [Next in Thread>