* [PATCH] scsi_debug: scale virtual_gb with sector_size properly
@ 2009-12-16 0:58 Jeff Mahoney
2009-12-17 19:57 ` Douglas Gilbert
0 siblings, 1 reply; 2+ messages in thread
From: Jeff Mahoney @ 2009-12-16 0:58 UTC (permalink / raw)
To: Linux SCSI Mailing List
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
The scsi_debug virtual_gb feature currently assumes that the sector size
is always 512 bytes and scales the capacity of virtual_gb based on that
assumption.
When passed a different sector size, the capacity of the disk is scaled
incorrectly higher than it should be.
For example, modprobe scsi_debug virtual_gb=1 sector_size=4096 results
in an 8 GB device.
This patch scales it appropriately. I chose to scale down the sectors-per-GB
value and then multiply so that if virtual_gb is changed to allow devices that
can max out sector_t, it will just work.
References: https://bugzilla.novell.com/show_bug.cgi?id=535939
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
- ---
drivers/scsi/scsi_debug.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
- --- a/drivers/scsi/scsi_debug.c
+++ b/drivers/scsi/scsi_debug.c
@@ -154,6 +154,7 @@ static int scsi_debug_virtual_gb = DEF_V
static int scsi_debug_fake_rw = DEF_FAKE_RW;
static int scsi_debug_vpd_use_hostno = DEF_VPD_USE_HOSTNO;
static int scsi_debug_sector_size = DEF_SECTOR_SIZE;
+static int scsi_debug_sector_size_bits = ilog2(DEF_SECTOR_SIZE);
static int scsi_debug_dix = DEF_DIX;
static int scsi_debug_dif = DEF_DIF;
static int scsi_debug_guard = DEF_GUARD;
@@ -895,9 +896,11 @@ static int resp_start_stop(struct scsi_c
static sector_t get_sdebug_capacity(void)
{
- - if (scsi_debug_virtual_gb > 0)
- - return 2048 * 1024 * (sector_t)scsi_debug_virtual_gb;
- - else
+ if (scsi_debug_virtual_gb > 0) {
+ sector_t sectors = 2 * 1024 * 1024;
+ sectors >>= (scsi_debug_sector_size_bits - 9);
+ return sectors * (sector_t)scsi_debug_virtual_gb;
+ } else
return sdebug_store_sectors;
}
@@ -2861,6 +2864,7 @@ static int __init scsi_debug_init(void)
case 1024:
case 2048:
case 4096:
+ scsi_debug_sector_size_bits = ilog2(scsi_debug_sector_size);
break;
default:
printk(KERN_ERR "scsi_debug_init: invalid sector_size %d\n",
- --
Jeff Mahoney
SuSE Labs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/
iEYEARECAAYFAksoMM8ACgkQLPWxlyuTD7JX5wCePXEPr9nsTescewxkCVnMKZ5e
GQ8An3sSVP1UhWAQ4O6LlKznvEWMg+tH
=CEfl
-----END PGP SIGNATURE-----
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] scsi_debug: scale virtual_gb with sector_size properly
2009-12-16 0:58 [PATCH] scsi_debug: scale virtual_gb with sector_size properly Jeff Mahoney
@ 2009-12-17 19:57 ` Douglas Gilbert
0 siblings, 0 replies; 2+ messages in thread
From: Douglas Gilbert @ 2009-12-17 19:57 UTC (permalink / raw)
To: Jeff Mahoney; +Cc: Linux SCSI Mailing List
Jeff Mahoney wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> The scsi_debug virtual_gb feature currently assumes that the sector size
> is always 512 bytes and scales the capacity of virtual_gb based on that
> assumption.
>
> When passed a different sector size, the capacity of the disk is scaled
> incorrectly higher than it should be.
>
> For example, modprobe scsi_debug virtual_gb=1 sector_size=4096 results
> in an 8 GB device.
>
> This patch scales it appropriately. I chose to scale down the sectors-per-GB
> value and then multiply so that if virtual_gb is changed to allow devices that
> can max out sector_t, it will just work.
>
> References: https://bugzilla.novell.com/show_bug.cgi?id=535939
>
> Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Acked-by: Douglas Gilbert <dgilbert@interlog.com>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2009-12-17 19:57 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-12-16 0:58 [PATCH] scsi_debug: scale virtual_gb with sector_size properly Jeff Mahoney
2009-12-17 19:57 ` Douglas Gilbert
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox