PERFORCE change 112379 for review

[Divacky Roman]

On Wed, Jan 03, 2007 at 02:59:19PM -0500, John Baldwin wrote:
> On Wednesday 03 January 2007 07:32, Divacky Roman wrote:
> > On Tue, Jan 02, 2007 at 11:29:54AM -0500, John Baldwin wrote:
> > > On Sunday 31 December 2006 18:53, Roman Divacky wrote:
> > > > http://perforce.freebsd.org/chv.cgi?CH=112379
> > > > 
> > > > Change 112379 by rdivacky at rdivacky_witten on 2006/12/31 23:53:30
> > > > 
> > > > 	Implement SETALL/GETALL IPC primitives. This fixes some LTP testcases.
> > > 
> > > Are you sure it's actually working?  I thought Linux semid structure was a
> > > different size, and GETALL and SETALL try to copyin and copyout data from
> > > userland without doing any conversions.
> > 
> > I only tested with the LTP tests and it fixed all cases there. 
> > 
> > the [SG]ETALL copies in/out u_char * nsems which looks quite OS independant.
> > 
> > someone reported that some program doesnt work because of missing 
> SETALL/GETALL
> > I think that if the one reportes that this patch fixes it we can consider it 
> ok.
> > (because we have 2 positive reports and 0 negative)
> > 
> > opinion?
> 
> Actually, it copies out an array of shorts so I do think it is ok and
> doesn't need any conversions.

by  "u_char * nsems" I meant exactly that :) the * stands for "multiply".

anyway - I'd wait for the one to confirm that it actually helps the program
and then Alexander can commit that.