On Mon, Dec 01, 2014 at 03:49:43PM +0000, Richard W.M. Jones wrote: > According to Documentation/arm64/booting.txt the dtb can occupy the > same page as the initrd. In fact qemu aligns it to 4K. However on a > kernel with 64K pages, you can end up freeing part of the device tree > when the initrd RAM is freed, causing bugs like this one: > > https://bugs.launchpad.net/qemu/+bug/1383857 > > Don't free partial first/last page when freeing initrd. > > Signed-off-by: Richard W.M. Jones <rjones@xxxxxxxxxx> > --- > arch/arm64/mm/init.c | 7 ++++--- > 1 file changed, 4 insertions(+), 3 deletions(-) > > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c > index 494297c..77bdc39 100644 > --- a/arch/arm64/mm/init.c > +++ b/arch/arm64/mm/init.c > @@ -335,11 +335,12 @@ void free_initrd_mem(unsigned long start, unsigned long end) > { > if (!keep_initrd) { > if (start == initrd_start) > - start = round_down(start, PAGE_SIZE); > + start = round_up(start, PAGE_SIZE); > if (end == initrd_end) > - end = round_up(end, PAGE_SIZE); > + end = round_down(end, PAGE_SIZE); > > - free_reserved_area((void *)start, (void *)end, 0, "initrd"); > + if (start < end) > + free_reserved_area((void *)start, (void *)end, 0, "initrd"); > } > } What about the equivalent fix for arch/arm/mm/init.c? Should probably post to linux-arm-kernel@xxxxxxxxxxxxxxxxxxx and linux-kernel@xxxxxxxxxxxxxxx drew > > -- > 2.1.0 > > _______________________________________________ > kvmarm mailing list > kvmarm@xxxxxxxxxxxxxxxxxxxxx > https://lists.cs.columbia.edu/mailman/listinfo/kvmarm _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
