[SeaBIOS] [PATCH RFC] use bootorder file supplied by qemu to order bootable devices
gleb at redhat.com
Mon Dec 13 16:02:43 CET 2010
On Mon, Dec 13, 2010 at 08:44:08AM -0500, Kevin O'Connor wrote:
> On Mon, Dec 13, 2010 at 03:12:51PM +0200, Gleb Natapov wrote:
> > On Thu, Dec 09, 2010 at 08:27:10PM -0500, Kevin O'Connor wrote:
> > > On Wed, Dec 08, 2010 at 02:00:50PM +0200, Gleb Natapov wrote:
> > > > With patch below (goes on top of bev per CD patch) I can specify boot
> > > > order from qemu command line between several NICs, CDROMs, floppy and
> > > > disk. With qemu command like this:
> > >
> > > Hi Gleb.
> > >
> > > I'm a bit behind right now, but I'll take a look this weekend.
> > >
> > Have you had a chance to look at it? If not yet then here is updated
> > version that fix one little bug with boot menu and choosing between
> > multiple BCVs:
> I did look at it, but didn't get a chance to reply. Here are my
> thoughts in brief:
> * I'm okay with the general layout (read bootorder at start, note
> priority when adding each entry)
> * The system seems to conflict with the sorting already done by
> add_bcv_internal() and add_ordered_drive() (which was added to
> handle threads adding drives in any order). The get_boot_priority()
> code you've added isn't right, because there's no guarentee that a
> drive will be added in a repeatable manor (you don't want a user
> without "bootorder" to see different drives being booted on
> different boots).
Good point. Will fix. Probably will require more invasive changes
> Also, it's not clear how this mechanism works
> with the boot_sequence variable.
IPLs are sorted by priority, IPL.bootorder is set to 0x87654321
(i.e try each IPL in order) and then boot_sequence is used
just like before the patch.
> * The strchr and romfile_name changes are fine and can go in now. The
> "bootorder" reading code looks fine too.
> * The ISA pci address is hardcoded (in floppy code), which isn't good.
Was too lazy :) Need to search for it dynamically.
> * I'm still unhappy about the parsing code.
Any specific comments that I can address?
> * I'd still like a way of designating an explicit rom (for coreboot
> only if necessary) - though I can add this on top.
> These comments are on the previous patch.
The last one just fixed a bug, so all of them still apply. Thanks.
More information about the SeaBIOS