Hi Ivan, Jay,
thanks for your comments.
On Thu, Dec 13, 2001 at 10:01:40PM +0300, Ivan Kokshaysky wrote:
> On Thu, Dec 13, 2001 at 01:02:18PM -0500, Jay Estabrook wrote:
> > 1. for all MIATAs, initially set the flag AVOID_DMA_OVER_8K_BOUNDARIES ;-}
> > 2. in "miata_map_irq()", clear that flag if returning IRQ 27, which
> > will only happen if on a MIATA GL
> > How does this sound?
> Sounds great - just a few lines of code.
Sounds good to me as well.
> Actually there is no need even for checking HWRPB - just add
> AVOID_DMA_OVER_8K_BOUNDARIES = 0xDEAD
> to miata_init_pci(). ;-)
So the workaround gets activated on all MIATAs, even the fixed ones (MIATA
GL). On the other hand, it does not hurt to use 16/17 DMA segments instead
of 4 segemnts for a 128k IDE DMA transfer.
> And the fix itself could be surrounded by
> #if defined(CONFIG_ALPHA_MIATA) || defined(CONFIG_ALPHA_GENERIC)
Sure. It already is. Except that I used CONFIG_ALPHA_CIA ||
CONFIG_ALPHA_GENERIC. (The CIA, because the test should have generally
worked on noards with CIAs. Who knows, maybe some early revision of 21171
shared the bug ... ? But as we can't detect it in an easy way, I'll just put
there CONFIG_ALPHA_PYXIS || ...GENERIC.)
-- Kurt Garloff <email@example.com> [Eindhoven, NL] Physics: Plasma simulations <K.Garloff@Phys.TUE.NL> [TU Eindhoven, NL] Linux: SCSI, Security <firstname.lastname@example.org> [SuSE Nuernberg, DE] (See mail header or public key servers for PGP2 and GPG public keys.)
This archive was generated by hypermail 2b29 : Sat Dec 15 2001 - 21:00:33 EST