Messages in this thread | | | Date | Sat, 25 Apr 1998 08:50:19 +1000 | From | Richard Gooch <> | Subject | Re: Lots of SCSI-disks, how?! |
| |
Mark H. Wood writes: > On Wed, 22 Apr 1998, Richard Gooch wrote: > > > David Woodhouse writes: > > > > > > Richard.Gooch@atnf.CSIRO.AU said: > > > > 16 SCSI discs is the limit because the Linux SCSI implementation > > > > supports 16 partitions per disc. With 8 bit minor numbers, you can > > > > only fit 16*16 (256) partitions and discs on a SCSI system. > > > > > > > One "solution" is to increase the minor number size, but that has > > > > problems with C library compatibilities. > > > > > > I thought glibc supported 64 bit dev_t. The filesystems might have problems, > > > though. > > > > There's still plently of people using libc 5 (and for good > > reasons!). I think there's even people using libc 4. If 2.3 were to > > introduce a 64 bit dev_t, it would stuff these people up, I think. > > While you could perhaps argue that libc 4 is so ancient that it > > doesn't have to be supported with kernel 2.3, I don't think libc 5 > > compatibility should be broken yet. > > Hey, I'm so cheap that I still have xv 3.00 (because it was the last > freeware version) and it was compiled against libc 4. Actually I've got > a fair amount of stuff that is still a.out and I've never found the > time to hunt down sources and rebuild it. Should I be clearing my > calendar for this task?
I rather we didn't break libc 4 compatibility if we can help it. I don't recall hearing anyone say that they have plans to break compatibility. As someone later pointed out, issues about kdev_t size can be minimised by using process personalities (i.e. insert structure rewriting code in certain syscalls). Doing stat on a device with major > 256 or minor > 256 would yield a wrapped device number, which will fould up some things.
However, see later comments in this thread about how we can survive with 16 bits device numbers but *still* support a roomful of discs: automagic generation of device numbers by devfs.
Regards,
Richard....
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |