Re: [LTP] [PATCH] IO_BITMAP_BITS definition in ioperm02

[Randy.Dunlap <rd...@xe...>]

On Sat, 10 Jun 2006 12:24:26 -0400 Mike Frysinger wrote:

> On Saturday 10 June 2006 11:56, Randy.Dunlap wrote:
> > On Sat, 10 Jun 2006 10:23:15 -0400 Mike Frysinger wrote:
> > > On Wednesday 17 May 2006 22:26, Mao, Bibo wrote:
> > > > IO_BITMAP_BITS value is different in different kernel version, from
> > > > view of kernel history, its value is changed from 2.6.8, but in 2.6.9
> > > > it is reverted to original value 1024. it is explained at website
> > > > http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.9.
> > >
> > > so you're saying that in 2.6.7 -> 2.6.8, the value was changed from 1024
> > > to something else, and then in 2.6.8 -> 2.6.9, it was changed back to
> > > 1024 ?
> >
> > It was increased to 65536 by this changeset on 2004-June-27:
> > http://linux.bkbits.net:8080/linux-2.6/cset@1.1757.49.11?nav=index.html|src
> >/|src/include|src/include/asm-i386|related/include/asm-i386/processor.h
> 
> ok ... so then the original patch proposed is incorrect and thus can be 
> ignored ?

Yes, that's how I read it (patch is not needed).

> linux-2.6.7 and older define IO_BITMAP_BITS as 1024
> 
> linux-2.6.8 and newer define IO_BITMAP_BITS as 65536
> 
> the ltp test uses the older value when running on kernels older than 2.6.8:
> if (tst_kvercmp(2,6,8) < 0) {
> 
> the proposed patch indicates that 2.6.9 also has IO_BITMAP_BITS as 1024:
> if (tst_kvercmp(2,6,8) < 0 || tst_kvercmp(2,6,9) == 0) {
> 
> but if we look at include/asm-i386/processor.h in the 2.6.{8,9,10} releases, 
> it is still set to 65536
> -mike

---
~Randy