[PATCH 0/3] x86: restore old GART alloc_coherent

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
From: FUJITA Tomonori
Date: Wednesday, September 24, 2008 - 4:48 am

This pachset is against tip/iommu.

What this patchset does is restoring old GART alloc_coherent behavior
(before the alloc_coherent rewrite):

http://lkml.org/lkml/2008/8/12/200

Currently, GART alloc_coherent tries to allocate pages with GFP_DMA32
for a device having dma_masks > 24bit < 32bits. If GART gets an
address that a device can't access to, GART tries to map the address
to a virtual I/O address that the device can access to.

But Andi pointed out, "The GART is somewhere in the 4GB range so you
cannot use it to map anything < 4GB. Also GART is pretty small."

http://lkml.org/lkml/2008/9/12/43

So it's possible that GART doesn't have virtual I/O address space that
a device can access to. The current behavior might not work for a
device having dma_masks > 24bit < 32bits. This patchset restores old
GART alloc_coherent behavior, which doesn't use GART hardware (if an
user doesn't enable force_iommu option).


--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
[PATCH 0/3] x86: restore old GART alloc_coherent, FUJITA Tomonori, (Wed Sep 24, 4:48 am)
[PATCH 1/3] x86: export pci-nommu's alloc_coherent, FUJITA Tomonori, (Wed Sep 24, 4:48 am)
[PATCH 3/3] x86: restore old GART alloc_coherent behavior, FUJITA Tomonori, (Wed Sep 24, 4:48 am)
Re: [PATCH 0/3] x86: restore old GART alloc_coherent, Joerg Roedel, (Wed Sep 24, 5:58 am)
Re: [PATCH 0/3] x86: restore old GART alloc_coherent, Ingo Molnar, (Thu Sep 25, 2:04 am)