lkml.org 
[lkml]   [2022]   [Sep]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v13 13/13] dm: add power-of-2 target for zoned devices with non power-of-2 zone sizes
From
>> +
>> + if (is_power_of_2(zone_size))
>> + DMWARN("%pg: underlying device has a power-of-2 number of sectors per zone",
>> + dmh->dev->bdev);
>> +
>> + dmh->zone_size = zone_size;
>> + dmh->zone_size_po2 = 1 << get_count_order_long(zone_size);
>> + dmh->zone_size_po2_shift = ilog2(dmh->zone_size_po2);
>> + dmh->zone_size_diff = dmh->zone_size_po2 - dmh->zone_size;
>> + ti->private = dmh;
>> + ti->max_io_len = dmh->zone_size_po2;
>> + dmh->nr_zones = npo2_zone_no(dmh, ti->len);
>> + ti->len = dmh->zone_size_po2 * dmh->nr_zones;
>> +
>> + return 0;
>> +}
>
> The above error paths need to unwind the references or any other
> resources acquired before failing. Please see other targets for how
> they handle sequencing of the needed operations (e.g. dm_put_device)
> in the error path by using gotos, etc.
>

Ok. That makes sense, and it should be pretty straight forward to do that.

>> +
>> +static void dm_po2z_io_hints(struct dm_target *ti, struct queue_limits *limits)
>> +{
>> + struct dm_po2z_target *dmh = ti->private;
>> +
>> + limits->chunk_sectors = dmh->zone_size_po2;
>> +}
>
> Are you certain you shouldn't at least be exposing a different
> logical_block_size to upper layers?
>
To be honest, I tested my patches in QEMU with 4k Logical block size and on
a device with 4k LBA size.

I did a quick test with 512B LBA size in QEMU, and I didn't see any
failures when I ran my normal test suite.

Do you see any problem with exposing the same LBA as the underlying device?

>> +
>> +static void dm_po2z_status(struct dm_target *ti, status_type_t type,
>> + unsigned int status_flags, char *result,
>> + unsigned int maxlen)
>> +{
>> + struct dm_po2z_target *dmh = ti->private;
>> + size_t sz = 0;
>> +
>> + switch (type) {
>> + case STATUSTYPE_INFO:
>> + DMEMIT("%s %lld", dmh->dev->name,
>> + (unsigned long long)dmh->zone_size_po2);
>> + break;
>
> Wouldn't it be worthwhile to expose the zone sectors (native npo2 vs
> simulated po2?) You merely roundup but never expose what you're using
> (unless I'm missing something about generic "zoned" device
> capabilities).
>

BLKREPORTZONE ioctl is typically used to get the zone information from a
zoned block device, which should expose the npo2 zone sectors(zone
capacity) in this case.

But I do see the value of exposing the dmh->zone_size instead of
dmh->zone_size_po2 as the latter can be easily calculated from the former
or it can be retrieved by reading the chunk_sectors. I will fix that up.


> Mike
>

\
 
 \ /
  Last update: 2022-09-14 21:17    [W:0.056 / U:0.808 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site