Marcelo,
Myself and Matt and previously others (searching the archive shows Eric
Sandeen complaining for quite a while) have noticed the following bug in
blkpg.c. You can see the most recent thread at:
http://groups.google.com/groups?hl=en&threadm=fa.fo6abnv.1shuppi%40ifi.uio.no&rnum=5&prev=/groups%3Fhl%3Den%26q%3DBLKGETSIZE64%26btnG%3DGoogle%2BSearch%26meta%3Dgroup%253Dfa.linux.kernel
Has this patch just missed inclusion or do you have a reason to leave
it out?
Tim
----- Forwarded message from Matt Domsch <[email protected]> -----
From: Matt Domsch <[email protected]>
To: <[email protected]>
Subject: BLKGETSIZE64 (bytes or sectors?)
Date: Thu, 17 Jan 2002 14:28:52 -0600 (CST)
X-Mailing-List: [email protected]
Is the BLKGETSIZE64 ioctl supposed to return the size of the device in
bytes (as the comment says, and is implemented in all places *except*
blkpg.c), or in sectors (as is implemented in blkpg.c since 2.4.15)?
It would seem that blkpg.c gets it wrong, that it should be in bytes.
Assuming that's the case, here's the patch to fix it against 2.4.18-pre4.
Thanks,
Matt
--
Matt Domsch
Sr. Software Engineer
Dell Linux Solutions http://www.dell.com/linux
#1 US Linux Server provider with 24.5% (IDC Dec 2001)
#2 Worldwide Linux Server provider with 18.2% (IDC Dec 2001)
--- linux-2.4.18-pre4/drivers/block/blkpg.c.orig Thu Jan 17 14:24:24 2002
+++ linux-2.4.18-pre4/drivers/block/blkpg.c Thu Jan 17 14:26:43 2002
@@ -247,7 +247,7 @@ int blk_ioctl(kdev_t dev, unsigned int c
if (cmd == BLKGETSIZE)
return put_user((unsigned long)ullval, (unsigned long *)arg);
else
- return put_user(ullval, (u64 *)arg);
+ return put_user((u64)ullval << 9 , (u64 *)arg);
#if 0
case BLKRRPART: /* Re-read partition tables */
if (!capable(CAP_SYS_ADMIN))
In bytes - the patch looks right to me