Message ID | 1501795433-982645-12-git-send-email-pasha.tatashin@oracle.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
(+ arm64 maintainers) Hi Pavel, On 3 August 2017 at 22:23, Pavel Tatashin <pasha.tatashin@oracle.com> wrote: > To optimize the performance of struct page initialization, > vmemmap_populate() will no longer zero memory. > > We must explicitly zero the memory that is allocated by vmemmap_populate() > for kasan, as this memory does not go through struct page initialization > path. > > Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com> > Reviewed-by: Steven Sistare <steven.sistare@oracle.com> > Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com> > Reviewed-by: Bob Picco <bob.picco@oracle.com> > --- > arch/arm64/mm/kasan_init.c | 32 ++++++++++++++++++++++++++++++++ > 1 file changed, 32 insertions(+) > > diff --git a/arch/arm64/mm/kasan_init.c b/arch/arm64/mm/kasan_init.c > index 81f03959a4ab..a57104bc54b8 100644 > --- a/arch/arm64/mm/kasan_init.c > +++ b/arch/arm64/mm/kasan_init.c > @@ -135,6 +135,31 @@ static void __init clear_pgds(unsigned long start, > set_pgd(pgd_offset_k(start), __pgd(0)); > } > > +/* > + * Memory that was allocated by vmemmap_populate is not zeroed, so we must > + * zero it here explicitly. > + */ > +static void > +zero_vemmap_populated_memory(void) Typo here: vemmap -> vmemmap > +{ > + struct memblock_region *reg; > + u64 start, end; > + > + for_each_memblock(memory, reg) { > + start = __phys_to_virt(reg->base); > + end = __phys_to_virt(reg->base + reg->size); > + > + if (start >= end) How would this ever be true? And why is it a stop condition? > + break; > + Are you missing a couple of kasan_mem_to_shadow() calls here? I can't believe your intention is to wipe all of DRAM. > + memset((void *)start, 0, end - start); > + } > + > + start = (u64)kasan_mem_to_shadow(_stext); > + end = (u64)kasan_mem_to_shadow(_end); > + memset((void *)start, 0, end - start); > +} > + > void __init kasan_init(void) > { > u64 kimg_shadow_start, kimg_shadow_end; > @@ -205,6 +230,13 @@ void __init kasan_init(void) > pfn_pte(sym_to_pfn(kasan_zero_page), PAGE_KERNEL_RO)); > > memset(kasan_zero_page, 0, PAGE_SIZE); > + > + /* > + * vmemmap_populate does not zero the memory, so we need to zero it > + * explicitly > + */ > + zero_vemmap_populated_memory(); > + > cpu_replace_ttbr1(lm_alias(swapper_pg_dir)); > > /* At this point kasan is fully initialized. Enable error messages */ > -- > 2.13.4 > KASAN uses vmemmap_populate as a convenience: kasan has nothing to do with vmemmap, but the function already existed and happened to do what KASAN requires. Given that that will no longer be the case, it would be far better to stop using vmemmap_populate altogether, and clone it into a KASAN specific version (with an appropriate name) with the zeroing folded into it.
Hi Ard, Thank you very much for reviewing this. I will fix the bug you found in the next iteration. >> +zero_vemmap_populated_memory(void) > > Typo here: vemmap -> vmemmap Yeap, will rename here, and in Intel variant. > >> +{ >> + struct memblock_region *reg; >> + u64 start, end; >> + >> + for_each_memblock(memory, reg) { >> + start = __phys_to_virt(reg->base); >> + end = __phys_to_virt(reg->base + reg->size); >> + >> + if (start >= end) > How would this ever be true? And why is it a stop condition? Yes this is a stop condition. Also look at the way kasan allocates its shadow memory in this file kasan_init(): 187 for_each_memblock(memory, reg) { 188 void *start = (void *)__phys_to_virt(reg->base); 189 void *end = (void *)__phys_to_virt(reg->base + reg->size); 190 191 if (start >= end) 192 break; ... 200 vmemmap_populate(...) >> + > > Are you missing a couple of kasan_mem_to_shadow() calls here? I can't > believe your intention is to wipe all of DRAM. True. Thank you for catching this bug. I have not really tested on arm, only compiled for sanity checking. Need to figure out how to configure qemu to run most generic arm code. I tested on x86 and sparc both real and qemu hardware. > > KASAN uses vmemmap_populate as a convenience: kasan has nothing to do > with vmemmap, but the function already existed and happened to do what > KASAN requires. > > Given that that will no longer be the case, it would be far better to > stop using vmemmap_populate altogether, and clone it into a KASAN > specific version (with an appropriate name) with the zeroing folded > into it. I agree, but this would be outside of the scope of this project. Pasha
diff --git a/arch/arm64/mm/kasan_init.c b/arch/arm64/mm/kasan_init.c index 81f03959a4ab..a57104bc54b8 100644 --- a/arch/arm64/mm/kasan_init.c +++ b/arch/arm64/mm/kasan_init.c @@ -135,6 +135,31 @@ static void __init clear_pgds(unsigned long start, set_pgd(pgd_offset_k(start), __pgd(0)); } +/* + * Memory that was allocated by vmemmap_populate is not zeroed, so we must + * zero it here explicitly. + */ +static void +zero_vemmap_populated_memory(void) +{ + struct memblock_region *reg; + u64 start, end; + + for_each_memblock(memory, reg) { + start = __phys_to_virt(reg->base); + end = __phys_to_virt(reg->base + reg->size); + + if (start >= end) + break; + + memset((void *)start, 0, end - start); + } + + start = (u64)kasan_mem_to_shadow(_stext); + end = (u64)kasan_mem_to_shadow(_end); + memset((void *)start, 0, end - start); +} + void __init kasan_init(void) { u64 kimg_shadow_start, kimg_shadow_end; @@ -205,6 +230,13 @@ void __init kasan_init(void) pfn_pte(sym_to_pfn(kasan_zero_page), PAGE_KERNEL_RO)); memset(kasan_zero_page, 0, PAGE_SIZE); + + /* + * vmemmap_populate does not zero the memory, so we need to zero it + * explicitly + */ + zero_vemmap_populated_memory(); + cpu_replace_ttbr1(lm_alias(swapper_pg_dir)); /* At this point kasan is fully initialized. Enable error messages */