RE: [PATCH] hpet: Fix struct_size() in kzalloc()

[David Laight]

> ...
> > > > -	hpetp = kzalloc(struct_size(hpetp, hp_dev, hdp->hd_nirqs - 1),
> > > > +	hpetp = kzalloc(struct_size(hpetp, hp_dev, hdp->hd_nirqs),
> > > >  			GFP_KERNEL);
> > > >
> > > >  	if (!hpetp)
> > > > --
> > >
> > > Yep, "char: hpet: Use flexible-array member" started causing random boot
> > > failures on mainline for me.  Tetsuo beat me to sending the fix.
> > >
> > > Only thing I'll add is that GFP_KERNEL can now fit on the previous line.
> >
> > Gustavo already sent a fix for this, I didn't realize it was causing
> > boot problems, I'll forward it on to Linus soon.
> 
> grep -r --include '*.c' 'struct_size.*- 1'
> gives 4 other matches.
> I bet they all have the same bug.
> There might be others that simple pattern doesn't find.

kvm_main.c is ok - although why it reallocates 1 entry smaller on
    delete is any bodies guess.
megaraid_sas_fusion.c is broken, header has seq[1]
   (and lots of __packed).
iavf_virtchnl.c is broken, header has list[1].
ice_sched.c is broken, header has generic[1].

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)