On 14/05/15 17:26, Luis R. Rodriguez wrote:
Tomi,
Any chance you can consider this series again? I realize you say you are
not familiar with MTRR but given its prevalence use on framebuffer device
drivers and since you are a framebuffer subsystem maintainer I think its
perhaps fair to ask one of you become familiar with this stuff, or ask
questions if there are any. If its of any help this patch which adds
ioremap_uc() already got merged and the logic to phase MTRR is described
there too:
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/?id=e4b6be33c28923d8cde53023e0888b1c5d1a9027
That patch won't even be needed on this series, this series happens to be
straight forward: if you just ioremap_nocache() once and use mtrr_add()
then conver over to ioremap_wc() and use arch_phys_wc_add(). If it already
used ioremap_wc() just convert over from mtrr_add() to arch_phys_wc_add()
Not sure if you will get an Acked-by by x86 folks on this series as they
are busy, but I'm adding bp and hpa just in case.
These are now queued for 4.2. Sorry it took so long. I've been away and
my bandwidth for the fbdev maintainership has been rather limited lately.
Tomi