Messages in this thread | | | Date | Fri, 12 Jun 1998 13:27:36 -0400 | From | "Theodore Y. Ts'o" <> | Subject | Re: OFFTOPIC: e2fsprogs and +2Gb partitions |
| |
From: alan@lxorguk.ukuu.org.uk (Alan Cox) Date: Fri, 12 Jun 1998 16:21:40 +0100 (BST)
> alan@lxorguk.ukuu.org.uk (Alan Cox) writes: > > > So why doesnt Ulrich add _LINUX_SOURCE for them ? > > Because there is a perfectly legal function lseek64.
Ok, can't argue with that. So mkfs etc should use open64/lseek64/close64 and be happy ?
Not and be compatible with libc5.
I wouldn't have had a problem with Ulrich removing llseek() from the prototypes if he had also removed llseek from libc. If lseek64() is the right interface to use (and correctly calls the Linux llseek system call), fine. Rename llseek to lseek64 and remove llseek altogether from libc.
But leaving llseek() in the library but removing it from the prototype, such that programs that use autoconf to determine whether or not llseek exists was just irresponsible. I had to add an e2fsprogs autoconf test to explicitly test for the presense of the function prototype (with comments grumbling about glibc) as a result of this change, and people who didn't know about this change and blithly recompiled e2fsprogs under glibc trashed their >2GB filesystems. There's no way I could have anticipated this when I released e2fsprogs 1.10, so I've been telling people that it was purely glibc's fault (or their fault for deciding to recompile e2fsprogs under glibc) that their filesystems were trashed.
- Ted
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |