From: Naohiro Aota <naohiro.aota@wdc.com>
To: Damien Le Moal <Damien.LeMoal@wdc.com>
Cc: Karel Zak <kzak@redhat.com>,
"util-linux@vger.kernel.org" <util-linux@vger.kernel.org>,
"linux-btrfs@vger.kernel.org" <linux-btrfs@vger.kernel.org>,
"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Subject: Re: [PATCH 3/3] blkid: support zone reset for wipefs
Date: Wed, 14 Apr 2021 10:23:51 +0900 [thread overview]
Message-ID: <20210414012351.eejntj3ribpg6nmr@naota-xeon> (raw)
In-Reply-To: <BL0PR04MB65141A1374C49EC06DDCCBB9E74F9@BL0PR04MB6514.namprd04.prod.outlook.com>
On Tue, Apr 13, 2021 at 10:47:00PM +0000, Damien Le Moal wrote:
> On 2021/04/14 7:11, Naohiro Aota wrote:
> > We cannot overwrite superblock magic in a sequential required zone. So,
> > wipefs cannot work as it is. Instead, this commit implements the wiping by
> > zone resetting.
> >
> > Zone resetting must be done only for a sequential write zone. This is
> > checked by is_conventional().
> >
> > Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
> > ---
> > libblkid/src/probe.c | 65 ++++++++++++++++++++++++++++++++++++++++----
> > 1 file changed, 59 insertions(+), 6 deletions(-)
> >
> > diff --git a/libblkid/src/probe.c b/libblkid/src/probe.c
> > index 102766e57aa0..7454a14bdfe6 100644
> > --- a/libblkid/src/probe.c
> > +++ b/libblkid/src/probe.c
> > @@ -107,6 +107,7 @@
> > #include <stdint.h>
> > #include <stdarg.h>
> > #include <limits.h>
> > +#include <stdbool.h>
> >
> > #include "blkidP.h"
> > #include "all-io.h"
> > @@ -1225,6 +1226,40 @@ int blkid_do_probe(blkid_probe pr)
> > return rc;
> > }
> >
> > +static int is_conventional(blkid_probe pr, uint64_t offset)
> > +{
> > + struct blk_zone_report *rep = NULL;
> > + size_t rep_size;
> > + bool conventional;
> > + int ret;
> > +
> > + if (!pr->zone_size)
> > + return 1;
> > +
> > + rep_size = sizeof(struct blk_zone_report) + sizeof(struct blk_zone);
> > + rep = malloc(rep_size);
> > + if (!rep)
> > + return -1;
> > +
> > + memset(rep, 0, rep_size);
>
> Use calloc() to get the zeroing done at the same time as allocation, may be ?
Will do.
> > + rep->sector = (offset & pr->zone_size) >> 9;
>
> Why the "& pr->zone_size" ? This seems wrong. To get the zone info of the zone
> containing offset, you can just have:
>
> rep->sector = offset >> 9;
>
> And if you want to align to the zone start, then:
>
> rep->sector = (offset & (~(pr->zone_size - 1))) >> 9;
Ouch. I was silly and missed this bug because it was only doing reset
on zone #0 and #1. I'll introduce "zone_mask" and do the correct
calculation.
> > + rep->nr_zones = 1;
> > + ret = ioctl(blkid_probe_get_fd(pr), BLKREPORTZONE, rep);
> > + if (ret) {
> > + free(rep);
> > + return -1;
> > + }
> > +
> > + if (rep->zones[0].type == BLK_ZONE_TYPE_CONVENTIONAL)
> > + ret = 1;
> > + else
> > + ret = 0;
> > +
> > + free(rep);
> > +
> > + return ret;
> > +}
> > +
> > /**
> > * blkid_do_wipe:
> > * @pr: prober
> > @@ -1264,6 +1299,7 @@ int blkid_do_wipe(blkid_probe pr, int dryrun)
> > const char *off = NULL;
> > size_t len = 0;
> > uint64_t offset, magoff;
> > + bool conventional;
> > char buf[BUFSIZ];
> > int fd, rc = 0;
> > struct blkid_chain *chn;
> > @@ -1299,6 +1335,11 @@ int blkid_do_wipe(blkid_probe pr, int dryrun)
> > if (len > sizeof(buf))
> > len = sizeof(buf);
> >
> > + rc = is_conventional(pr, offset);
> > + if (rc < 0)
> > + return rc;
> > + conventional = rc == 1;
> > +
> > DBG(LOWPROBE, ul_debug(
> > "do_wipe [offset=0x%"PRIx64" (%"PRIu64"), len=%zu, chain=%s, idx=%d, dryrun=%s]\n",
> > offset, offset, len, chn->driver->name, chn->idx, dryrun ? "yes" : "not"));
> > @@ -1306,13 +1347,25 @@ int blkid_do_wipe(blkid_probe pr, int dryrun)
> > if (lseek(fd, offset, SEEK_SET) == (off_t) -1)
> > return -1;
> >
> > - memset(buf, 0, len);
> > -
> > if (!dryrun && len) {
> > - /* wipen on device */
> > - if (write_all(fd, buf, len))
> > - return -1;
> > - fsync(fd);
> > + if (conventional) {
> > + memset(buf, 0, len);
> > +
> > + /* wipen on device */
> > + if (write_all(fd, buf, len))
> > + return -1;
> > + fsync(fd);
> > + } else {
> > + struct blk_zone_range range = {
> > + (offset & pr->zone_size) >> 9,
> > + pr->zone_size >> 9,
> > + };
>
> Please add the field names for clarity and fix the sector position (it looks
> very wrong):
>
> struct blk_zone_range range = {
> .sector = (offset & (~(pr->zone_size - 1))) >> 9,
> .nr_sectors = pr->zone_size >> 9,
> };
Will do the same.
> > +
> > + rc = ioctl(fd, BLKRESETZONE, &range);
> > + if (rc < 0)
> > + return -1;
> > + }
> > +
> > pr->flags &= ~BLKID_FL_MODIF_BUFF; /* be paranoid */
> >
> > return blkid_probe_step_back(pr);
> >
>
>
> --
> Damien Le Moal
> Western Digital Research
prev parent reply other threads:[~2021-04-14 1:24 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-13 22:10 [PATCH 0/3] implement zone-aware probing/wiping for zoned btrfs Naohiro Aota
2021-04-13 22:10 ` [PATCH 1/3] blkid: implement zone-aware probing Naohiro Aota
2021-04-13 22:38 ` Damien Le Moal
2021-04-14 1:20 ` Naohiro Aota
2021-04-13 22:10 ` [PATCH 2/3] blkid: add magic and probing for zoned btrfs Naohiro Aota
2021-04-13 22:10 ` [PATCH 3/3] blkid: support zone reset for wipefs Naohiro Aota
2021-04-13 22:47 ` Damien Le Moal
2021-04-14 1:23 ` Naohiro Aota [this message]
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=20210414012351.eejntj3ribpg6nmr@naota-xeon \
--to=naohiro.aota@wdc.com \
--cc=Damien.LeMoal@wdc.com \
--cc=Johannes.Thumshirn@wdc.com \
--cc=kzak@redhat.com \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=util-linux@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;
as well as URLs for NNTP newsgroup(s).