Commit d57c33c5daa4 (add generic fixmap.h) added (among other
similar things) set_fixmap_io to deal with early ioremap of devices.
More recently, commit bf4b558eba92 (arm64: add early_ioremap support)
converted the arm64 earlyprintk to use set_fixmap_io. A side effect of
this conversion is that my virtual machines have stopped booting when
I pass "earlyprintk=uart8250-8bit,0x3f8" to the guest kernel.
Turns out that the new earlyprintk code doesn't care at all about
sub-page offsets, and just assumes that the earlyprintk device will
be page-aligned. Obviously, that doesn't play well with the above example.
Further investigation shows that set_fixmap_io uses __set_fixmap instead
of __set_fixmap_offset. A fix is to introduce a set_fixmap_offset_io that
uses the latter, and to remove the superflous call to fix_to_virt
(which only returns the value that set_fixmap_io has already given us).
With this applied, my VMs are back in business. Tested on a Cortex-A57
platform with kvmtool as platform emulation.
Cc: Mark Salter <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
arch/arm64/kernel/early_printk.c | 6 ++----
include/asm-generic/fixmap.h | 3 +++
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/arch/arm64/kernel/early_printk.c b/arch/arm64/kernel/early_printk.c
index ffbbdde..2dc36d0 100644
--- a/arch/arm64/kernel/early_printk.c
+++ b/arch/arm64/kernel/early_printk.c
@@ -143,10 +143,8 @@ static int __init setup_early_printk(char *buf)
}
/* no options parsing yet */
- if (paddr) {
- set_fixmap_io(FIX_EARLYCON_MEM_BASE, paddr);
- early_base = (void __iomem *)fix_to_virt(FIX_EARLYCON_MEM_BASE);
- }
+ if (paddr)
+ early_base = (void __iomem *)set_fixmap_offset_io(FIX_EARLYCON_MEM_BASE, paddr);
printch = match->printch;
early_console = &early_console_dev;
diff --git a/include/asm-generic/fixmap.h b/include/asm-generic/fixmap.h
index 5a64ca4..f23174f 100644
--- a/include/asm-generic/fixmap.h
+++ b/include/asm-generic/fixmap.h
@@ -93,5 +93,8 @@ static inline unsigned long virt_to_fix(const unsigned long vaddr)
#define set_fixmap_io(idx, phys) \
__set_fixmap(idx, phys, FIXMAP_PAGE_IO)
+#define set_fixmap_offset_io(idx, phys) \
+ __set_fixmap_offset(idx, phys, FIXMAP_PAGE_IO)
+
#endif /* __ASSEMBLY__ */
#endif /* __ASM_GENERIC_FIXMAP_H */
--
1.8.3.4
On Mon, 2014-04-28 at 19:50 +0100, Marc Zyngier wrote:
> Commit d57c33c5daa4 (add generic fixmap.h) added (among other
> similar things) set_fixmap_io to deal with early ioremap of devices.
>
> More recently, commit bf4b558eba92 (arm64: add early_ioremap support)
> converted the arm64 earlyprintk to use set_fixmap_io. A side effect of
> this conversion is that my virtual machines have stopped booting when
> I pass "earlyprintk=uart8250-8bit,0x3f8" to the guest kernel.
>
> Turns out that the new earlyprintk code doesn't care at all about
> sub-page offsets, and just assumes that the earlyprintk device will
> be page-aligned. Obviously, that doesn't play well with the above example.
>
> Further investigation shows that set_fixmap_io uses __set_fixmap instead
> of __set_fixmap_offset. A fix is to introduce a set_fixmap_offset_io that
> uses the latter, and to remove the superflous call to fix_to_virt
> (which only returns the value that set_fixmap_io has already given us).
>
> With this applied, my VMs are back in business. Tested on a Cortex-A57
> platform with kvmtool as platform emulation.
>
> Cc: Mark Salter <[email protected]>
> Cc: Catalin Marinas <[email protected]>
> Cc: Will Deacon <[email protected]>
> Signed-off-by: Marc Zyngier <[email protected]>
> ---
Acked-by: Mark Salter <[email protected]>
On Mon, Apr 28, 2014 at 1:50 PM, Marc Zyngier <[email protected]> wrote:
> Commit d57c33c5daa4 (add generic fixmap.h) added (among other
> similar things) set_fixmap_io to deal with early ioremap of devices.
>
> More recently, commit bf4b558eba92 (arm64: add early_ioremap support)
> converted the arm64 earlyprintk to use set_fixmap_io. A side effect of
> this conversion is that my virtual machines have stopped booting when
> I pass "earlyprintk=uart8250-8bit,0x3f8" to the guest kernel.
>
> Turns out that the new earlyprintk code doesn't care at all about
> sub-page offsets, and just assumes that the earlyprintk device will
> be page-aligned. Obviously, that doesn't play well with the above example.
>
> Further investigation shows that set_fixmap_io uses __set_fixmap instead
> of __set_fixmap_offset. A fix is to introduce a set_fixmap_offset_io that
> uses the latter, and to remove the superflous call to fix_to_virt
> (which only returns the value that set_fixmap_io has already given us).
>
> With this applied, my VMs are back in business. Tested on a Cortex-A57
> platform with kvmtool as platform emulation.
>
> Cc: Mark Salter <[email protected]>
> Cc: Catalin Marinas <[email protected]>
> Cc: Will Deacon <[email protected]>
> Signed-off-by: Marc Zyngier <[email protected]>
> ---
> arch/arm64/kernel/early_printk.c | 6 ++----
> include/asm-generic/fixmap.h | 3 +++
> 2 files changed, 5 insertions(+), 4 deletions(-)
This will be fixed already in 3.16 with my earlycon series[1] if this
is not taken for 3.15.
Probably still a good addition for fixmap.
Rob
[1] https://lkml.org/lkml/2014/4/18/573
On Tue, Apr 29, 2014 at 12:51:18AM +0100, Rob Herring wrote:
> On Mon, Apr 28, 2014 at 1:50 PM, Marc Zyngier <[email protected]> wrote:
> > Commit d57c33c5daa4 (add generic fixmap.h) added (among other
> > similar things) set_fixmap_io to deal with early ioremap of devices.
> >
> > More recently, commit bf4b558eba92 (arm64: add early_ioremap support)
> > converted the arm64 earlyprintk to use set_fixmap_io. A side effect of
> > this conversion is that my virtual machines have stopped booting when
> > I pass "earlyprintk=uart8250-8bit,0x3f8" to the guest kernel.
> >
> > Turns out that the new earlyprintk code doesn't care at all about
> > sub-page offsets, and just assumes that the earlyprintk device will
> > be page-aligned. Obviously, that doesn't play well with the above example.
> >
> > Further investigation shows that set_fixmap_io uses __set_fixmap instead
> > of __set_fixmap_offset. A fix is to introduce a set_fixmap_offset_io that
> > uses the latter, and to remove the superflous call to fix_to_virt
> > (which only returns the value that set_fixmap_io has already given us).
> >
> > With this applied, my VMs are back in business. Tested on a Cortex-A57
> > platform with kvmtool as platform emulation.
> >
> > Cc: Mark Salter <[email protected]>
> > Cc: Catalin Marinas <[email protected]>
> > Cc: Will Deacon <[email protected]>
> > Signed-off-by: Marc Zyngier <[email protected]>
> > ---
> > arch/arm64/kernel/early_printk.c | 6 ++----
> > include/asm-generic/fixmap.h | 3 +++
> > 2 files changed, 5 insertions(+), 4 deletions(-)
>
> This will be fixed already in 3.16 with my earlycon series[1] if this
> is not taken for 3.15.
I'd like to take this for 3.15 as it's currently broken. I just need
another ack from Arnd on the generic header change.
--
Catalin