On Thu, 2007-03-08 at 15:43 +0100, Jim Meyering wrote: > Here's a fix for the first memory overrun bug I found: > > aix.c: Avoid memory overrun. Don't assume logical sector size <= 512B > * libparted/labels/aix.c (aix_probe): Return 0 if the > sector size is larger than our AixLabel size. > (aix_clobber): Rather than PED_ASSERT'ing that aix_probe returns 1, > simply return 0 if aix_probe returns fails. > > diff --git a/libparted/labels/aix.c b/libparted/labels/aix.c > index a16ead4..9e2a7bb 100644 > --- a/libparted/labels/aix.c > +++ b/libparted/labels/aix.c > @@ -48,6 +48,8 @@ aix_probe (const PedDevice *dev) > AixLabel label; > > PED_ASSERT (dev != NULL, return 0); > + if (sizeof (AixLabel) < dev->sector_size) > + return 0; > > if (!ped_device_read (dev, &label, 0, 1)) > return 0; > @@ -65,7 +67,8 @@ aix_clobber (PedDevice* dev) > AixLabel label; > > PED_ASSERT (dev != NULL, return 0); > - PED_ASSERT (aix_probe (dev), return 0); > + if (!aix_probe (dev)) > + return 0; > > if (!ped_device_read (dev, &label, 0, 1)) > return 0; > ----------------------------------------------- > > The above is similar to what's done in dos.c's msdos_probe: > > if (dev->sector_size != 512) > return 0; > > Is it possible to have a DOS or AIX partition on a CDROM with 2048-byte > logical sectors? (I have no idea) If so, then it might make sense to
DOS, yes. AIX...no clue. > do what rdb.c's amiga_probe does: > > if ((rdb=RDSK(ped_malloc(dev->sector_size)))==NULL) > return 0; > > i.e., rather than simply returning when dev->sector_size is too large > or != 512, just allocate a buffer of the required size and use that, > rather than using the fixed-size one on the stack. I like the approach of allocating the required buffer size instead of using fixed-size buffers. Either solution is fine, but I would prefer we go with the ped_malloc() approach that's used in amiga_probe so we are consistent throughout libparted. -- David Cantrell <[EMAIL PROTECTED]> Red Hat / Westford, MA
signature.asc
Description: This is a digitally signed message part
_______________________________________________ parted-devel mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/parted-devel
