Messages in this thread |  | | Subject | Re: [PATCH] mm, kasan: don't poison boot memory | From | George Kennedy <> | Date | Mon, 22 Feb 2021 10:13:24 -0500 |
| |
On 2/22/2021 4:52 AM, David Hildenbrand wrote: > On 20.02.21 00:04, George Kennedy wrote: >> >> >> On 2/19/2021 11:45 AM, George Kennedy wrote: >>> >>> >>> On 2/18/2021 7:09 PM, Andrey Konovalov wrote: >>>> On Fri, Feb 19, 2021 at 1:06 AM George Kennedy >>>> <george.kennedy@oracle.com> wrote: >>>>> >>>>> >>>>> On 2/18/2021 3:55 AM, David Hildenbrand wrote: >>>>>> On 17.02.21 21:56, Andrey Konovalov wrote: >>>>>>> During boot, all non-reserved memblock memory is exposed to the >>>>>>> buddy >>>>>>> allocator. Poisoning all that memory with KASAN lengthens boot >>>>>>> time, >>>>>>> especially on systems with large amount of RAM. This patch makes >>>>>>> page_alloc to not call kasan_free_pages() on all new memory. >>>>>>> >>>>>>> __free_pages_core() is used when exposing fresh memory during >>>>>>> system >>>>>>> boot and when onlining memory during hotplug. This patch adds a new >>>>>>> FPI_SKIP_KASAN_POISON flag and passes it to __free_pages_ok() >>>>>>> through >>>>>>> free_pages_prepare() from __free_pages_core(). >>>>>>> >>>>>>> This has little impact on KASAN memory tracking. >>>>>>> >>>>>>> Assuming that there are no references to newly exposed pages >>>>>>> before they >>>>>>> are ever allocated, there won't be any intended (but buggy) >>>>>>> accesses to >>>>>>> that memory that KASAN would normally detect. >>>>>>> >>>>>>> However, with this patch, KASAN stops detecting wild and large >>>>>>> out-of-bounds accesses that happen to land on a fresh memory page >>>>>>> that >>>>>>> was never allocated. This is taken as an acceptable trade-off. >>>>>>> >>>>>>> All memory allocated normally when the boot is over keeps getting >>>>>>> poisoned as usual. >>>>>>> >>>>>>> Signed-off-by: Andrey Konovalov <andreyknvl@google.com> >>>>>>> Change-Id: Iae6b1e4bb8216955ffc14af255a7eaaa6f35324d >>>>>> Not sure this is the right thing to do, see >>>>>> >>>>>> https://lkml.kernel.org/r/bcf8925d-0949-3fe1-baa8-cc536c529860@oracle.com >>>>>> >>>>>> >>>>>> >>>>>> Reversing the order in which memory gets allocated + used during >>>>>> boot >>>>>> (in a patch by me) might have revealed an invalid memory access >>>>>> during >>>>>> boot. >>>>>> >>>>>> I suspect that that issue would no longer get detected with your >>>>>> patch, as the invalid memory access would simply not get detected. >>>>>> Now, I cannot prove that :) >>>>> Since David's patch we're having trouble with the iBFT ACPI table, >>>>> which >>>>> is mapped in via kmap() - see acpi_map() in "drivers/acpi/osl.c". >>>>> KASAN >>>>> detects that it is being used after free when ibft_init() accesses >>>>> the >>>>> iBFT table, but as of yet we can't find where it get's freed (we've >>>>> instrumented calls to kunmap()). >>>> Maybe it doesn't get freed, but what you see is a wild or a large >>>> out-of-bounds access. Since KASAN marks all memory as freed during the >>>> memblock->page_alloc transition, such bugs can manifest as >>>> use-after-frees. >>> >>> It gets freed and re-used. By the time the iBFT table is accessed by >>> ibft_init() the page has been over-written. >>> >>> Setting page flags like the following before the call to kmap() >>> prevents the iBFT table page from being freed: >> >> Cleaned up version: >> >> diff --git a/drivers/acpi/osl.c b/drivers/acpi/osl.c >> index 0418feb..8f0a8e7 100644 >> --- a/drivers/acpi/osl.c >> +++ b/drivers/acpi/osl.c >> @@ -287,9 +287,12 @@ static void __iomem *acpi_map(acpi_physical_address >> pg_off, unsigned long pg_sz) >> >> pfn = pg_off >> PAGE_SHIFT; >> if (should_use_kmap(pfn)) { >> + struct page *page = pfn_to_page(pfn); >> + >> if (pg_sz > PAGE_SIZE) >> return NULL; >> - return (void __iomem __force *)kmap(pfn_to_page(pfn)); >> + SetPageReserved(page); >> + return (void __iomem __force *)kmap(page); >> } else >> return acpi_os_ioremap(pg_off, pg_sz); >> } >> @@ -299,9 +302,12 @@ static void acpi_unmap(acpi_physical_address >> pg_off, void __iomem *vaddr) >> unsigned long pfn; >> >> pfn = pg_off >> PAGE_SHIFT; >> - if (should_use_kmap(pfn)) >> - kunmap(pfn_to_page(pfn)); >> - else >> + if (should_use_kmap(pfn)) { >> + struct page *page = pfn_to_page(pfn); >> + >> + ClearPageReserved(page); >> + kunmap(page); >> + } else >> iounmap(vaddr); >> } >> >> David, the above works, but wondering why it is now necessary. kunmap() >> is not hit. What other ways could a page mapped via kmap() be unmapped? >> > > Let me look into the code ... I have little experience with ACPI > details, so bear with me. > > I assume that acpi_map()/acpi_unmap() map some firmware blob that is > provided via firmware/bios/... to us. > > should_use_kmap() tells us whether > a) we have a "struct page" and should kmap() that one > b) we don't have a "struct page" and should ioremap. > > As it is a blob, the firmware should always reserve that memory region > via memblock (e.g., memblock_reserve()), such that we either > 1) don't create a memmap ("struct page") at all (-> case b) ) > 2) if we have to create e memmap, we mark the page PG_reserved and > *never* expose it to the buddy (-> case a) ) > > > Are you telling me that in this case we might have a memmap for the HW > blob that is *not* PG_reserved? In that case it most probably got > exposed to the buddy where it can happily get allocated/freed. > > The latent BUG would be that that blob gets exposed to the system like > ordinary RAM, and not reserved via memblock early during boot. > Assuming that blob has a low physical address, with my patch it will > get allocated/used a lot earlier - which would mean we trigger this > latent BUG now more easily. > > There have been similar latent BUGs on ARM boards that my patch > discovered where special RAM regions did not get marked as reserved > via the device tree properly. > > Now, this is just a wild guess :) Can you dump the page when mapping > (before PageReserved()) and when unmapping, to see what the state of > that memmap is?
Thank you David for the explanation and your help on this,
dump_page() before PageReserved and before kmap() in the above patch:
[ 1.116480] ACPI: Core revision 20201113 [ 1.117628] XXX acpi_map: about to call kmap()... [ 1.118561] page:ffffea0002f914c0 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0xbe453 [ 1.120381] flags: 0xfffffc0000000() [ 1.121116] raw: 000fffffc0000000 ffffea0002f914c8 ffffea0002f914c8 0000000000000000 [ 1.122638] raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000 [ 1.124146] page dumped because: acpi_map pre SetPageReserved
I also added dump_page() before unmapping, but it is not hit. The following for the same pfn now shows up I believe as a result of setting PageReserved:
[ 28.098208] BUG: Bad page state in process modprobe pfn:be453 [ 28.098394] page:ffffea0002f914c0 refcount:0 mapcount:0 mapping:0000000000000000 index:0x1 pfn:0xbe453 [ 28.098394] flags: 0xfffffc0001000(reserved) [ 28.098394] raw: 000fffffc0001000 dead000000000100 dead000000000122 0000000000000000 [ 28.098394] raw: 0000000000000001 0000000000000000 00000000ffffffff 0000000000000000 [ 28.098394] page dumped because: PAGE_FLAGS_CHECK_AT_PREP flag(s) set [ 28.098394] page_owner info is not present (never set?) [ 28.098394] Modules linked in: [ 28.098394] CPU: 2 PID: 204 Comm: modprobe Not tainted 5.11.0-3dbd5e3 #66 [ 28.098394] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 0.0.0 02/06/2015 [ 28.098394] Call Trace: [ 28.098394] dump_stack+0xdb/0x120 [ 28.098394] bad_page.cold.108+0xc6/0xcb [ 28.098394] check_new_page_bad+0x47/0xa0 [ 28.098394] get_page_from_freelist+0x30cd/0x5730 [ 28.098394] ? __isolate_free_page+0x4f0/0x4f0 [ 28.098394] ? init_object+0x7e/0x90 [ 28.098394] __alloc_pages_nodemask+0x2d8/0x650 [ 28.098394] ? write_comp_data+0x2f/0x90 [ 28.098394] ? __alloc_pages_slowpath.constprop.103+0x2110/0x2110 [ 28.098394] ? __sanitizer_cov_trace_pc+0x21/0x50 [ 28.098394] alloc_pages_vma+0xe2/0x560 [ 28.098394] do_fault+0x194/0x12c0 [ 28.098394] ? write_comp_data+0x2f/0x90 [ 28.098394] __handle_mm_fault+0x1650/0x26c0 [ 28.098394] ? copy_page_range+0x1350/0x1350 [ 28.098394] ? write_comp_data+0x2f/0x90 [ 28.098394] ? write_comp_data+0x2f/0x90 [ 28.098394] handle_mm_fault+0x1f9/0x810 [ 28.098394] ? write_comp_data+0x2f/0x90 [ 28.098394] do_user_addr_fault+0x6f7/0xca0 [ 28.098394] exc_page_fault+0xaf/0x1a0 [ 28.098394] asm_exc_page_fault+0x1e/0x30 [ 28.098394] RIP: 0010:__clear_user+0x30/0x60
What would be the correct way to reserve the page so that the above would not be hit?
BTW, this is running with Konrad's patch that pairs acpi_get_table & acpi_put_table for the iBFT table which should result in an eventual call to acpi_unmap() and kunmap(), though that does not occur. Could be a possible acpi page refcount issue that will have to be looked into.
George
|  |