ARM: mm: Fix for highmem on machines with more than 14 cores
RFC PATCH 1/1 ARM: mm: Fix for highmem on machines with more than 14 cores
Hi,
While working of platform support for the LSI AXM5516 SoC (a 16 x A15 device) we ran into this limitation in fixmap.h where there is not enough room for the per-cpu pages if NR_CPUS is set to 16.
The following patch avoids the overlap by lowering the base address according to the values of NR_CPUS and KM_TYPE_NR. Is this the way to handle this?
Thanks,
Anders
Anders Berg (1):
ARM: mm: Fix for highmem on machines with more than 14 cores
arch/arm/include/asm/fixmap.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
Signature · 1 line
1.8.3.2
RFC PATCH 1/1 ARM: mm: Fix for highmem on machines with more than 14 cores
The fixmap area was hard coded to 224 pages which was only enough for 14 CPU:s. Now the area adapts to the configured number of CPU:s (NR_CPUS). We still have a hard limit of 30 CPU:s since we only have a single pte-chunk (512 pte:s).
--- arch/arm/include/asm/fixmap.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
arch/arm/include/asm/fixmap.h +3 -3
@@ -7,15 +7,15 @@* On ARM we already have well known fixed virtual addresses imposed by * the architecture such as the vector page which is located at 0xffff0000, * therefore a second level page table is already allocated covering -* 0xfff00000 upwards. +* 0xffe00000 upwards. * * The cache flushing code in proc-xscale.S uses the virtual area between * 0xfffe0000 and 0xfffeffff. */ -#define FIXADDR_START 0xfff00000UL +#define FIXADDR_SIZE ((KM_TYPE_NR * NR_CPUS) << PAGE_SHIFT)#define FIXADDR_TOP 0xfffe0000UL -#define FIXADDR_SIZE (FIXADDR_TOP - FIXADDR_START) +#define FIXADDR_START (FIXADDR_TOP - FIXADDR_SIZE)#define FIX_KMAP_BEGIN 0#define FIX_KMAP_END (FIXADDR_SIZE >> PAGE_SHIFT)
Signature · 1 line
1.8.3.2
Quoted 6 lines · Anders Berg
The fixmap area was hard coded to 224 pages which was only enough for 14 CPU:s. Now the area adapts to the configured number of CPU:s (
NR_CPUS). We still have a hard limit of 30 CPU:s since we only have a single pte-chunk (512 pte:s).Signed-off-by Anders Berg <[email protected]>Signed-off-by Fredrik Markstr?m <[email protected]>
This is under discussion already. Please have a look at this thread:
http://news.gmane.org/group/gmane.linux.kernel/thread=1672068/
Nicolas