|
@mjg59 | |||||
|
My patch to disable PCI device DMA in early boot to avoid gaps in IOMMU coverage just got merged to mainline, so here's a writeup of it: mjg59.dreamwidth.org/54433.html
|
||||||
|
||||||
|
Jeff Waugh
@jdub
|
29. sij |
|
Should coreboot/u-boot do this as early as possible too?
|
||
|
|
||
|
Matthew Garrett
@mjg59
|
29. sij |
|
It's sort of complicated - coreboot itself is what would have to enable it, but coreboot has no idea what payload is going to be launched (eg, if it's SeaBIOS then something's going to have to remove the state again)
|
||
|
|
||
|
Shift
@Shiftreduce
|
28. sij |
|
congrats!
|
||
|
|
||
|
Eloy
@rfc1087
|
28. sij |
|
Awesome! Thanks :)
|
||
|
|
||
|
Pedro Ribeiro
@pedrib1337
|
29. sij |
|
Now if only Intel fixed their iommu - on some systems it is hopelessly broken and causes lots of crashes.
|
||
|
|
||
|
spiffycell
@spiffycell
|
29. sij |
|
This is awesome - thank you for this!
|
||
|
|
||