Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Thu, Jan 21, 2021 at 12:30 PM Vincenzo Frascino wrote: > > Hi Andrey, > > On 1/19/21 8:56 PM, Andrey Konovalov wrote: > >> return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); > > Do we need is_vmalloc_addr()? As we don't yet have vmalloc support for > > HW_TAGS. > > It is not necessary but it does not hurt, since we are going to add vmalloc > anyway at some point, I would keep it here. OK, let's keep it.
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
Hi Andrey, On 1/19/21 8:56 PM, Andrey Konovalov wrote: >> return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); > Do we need is_vmalloc_addr()? As we don't yet have vmalloc support for > HW_TAGS. It is not necessary but it does not hurt, since we are going to add vmalloc anyway at some point, I would keep it here. -- Regards, Vincenzo
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Wed, Jan 20, 2021 at 04:16:02PM +, Vincenzo Frascino wrote: > On 1/20/21 4:04 PM, Catalin Marinas wrote: > > On Tue, Jan 19, 2021 at 08:35:49PM +, Vincenzo Frascino wrote: > >> On 1/19/21 6:52 PM, Catalin Marinas wrote: > >>> On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: > On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > wrote: > > > > With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > > the address passed as a parameter. > > > > Add a comment to make sure that the preconditions to the function are > > explicitly clarified. > > > > Note: An invalid address (e.g. NULL pointer address) passed to the > > function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > > > > Cc: Andrey Ryabinin > > Cc: Alexander Potapenko > > Cc: Dmitry Vyukov > > Cc: Leon Romanovsky > > Cc: Andrey Konovalov > > Signed-off-by: Vincenzo Frascino > > --- > > mm/kasan/report.c | 11 +++ > > 1 file changed, 11 insertions(+) > > > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c > > index c0fb21797550..2485b585004d 100644 > > --- a/mm/kasan/report.c > > +++ b/mm/kasan/report.c > > @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > > size_t size, bool is_write, > > end_report(); > > } > > > > +/** > > + * kasan_report - report kasan fault details > > + * @addr: valid address of the allocation where the tag fault was > > detected > > + * @size: size of the allocation where the tag fault was detected > > + * @is_write: the instruction that caused the fault was a read or > > write? > > + * @ip: pointer to the instruction that cause the fault > > + * > > + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() > > dereferences > > + * the address to access the tags, hence it must be valid at this > > point in > > + * order to not cause a kernel panic. > > + */ > > It doesn't dereference the address, it just checks the tags, right? > > Ideally, kasan_report() should survive that with HW_TAGS like with the > other modes. The reason it doesn't is probably because of a blank > addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > guess we should somehow check that the memory comes from page_alloc or > kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > instruction to check whether the memory has tags? > >>> > >>> There isn't an architected way to probe whether a memory location has a > >>> VA->PA mapping. The tags are addressed by PA but you can't reach them if > >>> you get a page fault on the VA. So we either document the kasan_report() > >>> preconditions or, as you suggest, update addr_has_metadata() for the > >>> HW_TAGS case. Something like: > >>> > >>> return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > >>> > >> > >> This seems not working on arm64 because according to virt_addr_valid 0 is a > >> valid virtual address, in fact: > >> > >> __is_lm_address(0) == true && pfn_valid(virt_to_pfn(0)) == true. > > > > Ah, so __is_lm_address(0) is true. Maybe we should improve this since > > virt_to_pfn(0) doesn't make much sense. > > How do you propose to improve it? Check that it's actually a kernel address starting at PAGE_OFFSET. The current __is_lm_address() check just masks out the top 12 bits but if they were 0, this still yields a true result. Maybe extending the current definition as: #define __is_lm_address(addr) ((u64)(addr) >= PAGE_OFFSET && \ ((u64)(addr) & ~PAGE_OFFSET) < (PAGE_END - PAGE_OFFSET)) Which basically means: #define __is_lm_address(addr) ((u64)(addr) >= PAGE_OFFSET && \ (u64)(addr) < PAGE_END) I think we could write the above as: #define __is_lm_address(addr) (((u64)(addr) ^ PAGE_OFFSET) < (PAGE_END - PAGE_OFFSET)) This way we catch any 0 bits in the top 12 (or 16 with a 48-bit VA configuration). -- Catalin
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On 1/20/21 4:04 PM, Catalin Marinas wrote: > On Tue, Jan 19, 2021 at 08:35:49PM +, Vincenzo Frascino wrote: >> On 1/19/21 6:52 PM, Catalin Marinas wrote: >>> On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino wrote: > > With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > the address passed as a parameter. > > Add a comment to make sure that the preconditions to the function are > explicitly clarified. > > Note: An invalid address (e.g. NULL pointer address) passed to the > function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > > Cc: Andrey Ryabinin > Cc: Alexander Potapenko > Cc: Dmitry Vyukov > Cc: Leon Romanovsky > Cc: Andrey Konovalov > Signed-off-by: Vincenzo Frascino > --- > mm/kasan/report.c | 11 +++ > 1 file changed, 11 insertions(+) > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c > index c0fb21797550..2485b585004d 100644 > --- a/mm/kasan/report.c > +++ b/mm/kasan/report.c > @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > size_t size, bool is_write, > end_report(); > } > > +/** > + * kasan_report - report kasan fault details > + * @addr: valid address of the allocation where the tag fault was > detected > + * @size: size of the allocation where the tag fault was detected > + * @is_write: the instruction that caused the fault was a read or write? > + * @ip: pointer to the instruction that cause the fault > + * > + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > + * the address to access the tags, hence it must be valid at this point > in > + * order to not cause a kernel panic. > + */ It doesn't dereference the address, it just checks the tags, right? Ideally, kasan_report() should survive that with HW_TAGS like with the other modes. The reason it doesn't is probably because of a blank addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I guess we should somehow check that the memory comes from page_alloc or kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm instruction to check whether the memory has tags? >>> >>> There isn't an architected way to probe whether a memory location has a >>> VA->PA mapping. The tags are addressed by PA but you can't reach them if >>> you get a page fault on the VA. So we either document the kasan_report() >>> preconditions or, as you suggest, update addr_has_metadata() for the >>> HW_TAGS case. Something like: >>> >>> return is_vmalloc_addr(virt) || virt_addr_valid(virt)); >>> >> >> This seems not working on arm64 because according to virt_addr_valid 0 is a >> valid virtual address, in fact: >> >> __is_lm_address(0) == true && pfn_valid(virt_to_pfn(0)) == true. > > Ah, so __is_lm_address(0) is true. Maybe we should improve this since > virt_to_pfn(0) doesn't make much sense. > How do you propose to improve it? >> An option could be to make an exception for virtual address 0 in >> addr_has_metadata() something like: >> >> static inline bool addr_has_metadata(const void *addr) >> { >> if ((u64)addr == 0) >> return false; >> >> return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); >> } > > As Andrey replied, passing a non-zero small value would still be > incorrectly detected as valid. > I would like to remove the check completely and have virt_addr_valid(addr) to return the right thing if possible. I admit, yesterday evening I did not thing it through completely before posting this code that had the sole purpose to open the discussion. I agree in principle on what Andrey said as well (addr < PAGE_SIZE). -- Regards, Vincenzo
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 08:35:49PM +, Vincenzo Frascino wrote: > On 1/19/21 6:52 PM, Catalin Marinas wrote: > > On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: > >> On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > >> wrote: > >>> > >>> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > >>> the address passed as a parameter. > >>> > >>> Add a comment to make sure that the preconditions to the function are > >>> explicitly clarified. > >>> > >>> Note: An invalid address (e.g. NULL pointer address) passed to the > >>> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > >>> > >>> Cc: Andrey Ryabinin > >>> Cc: Alexander Potapenko > >>> Cc: Dmitry Vyukov > >>> Cc: Leon Romanovsky > >>> Cc: Andrey Konovalov > >>> Signed-off-by: Vincenzo Frascino > >>> --- > >>> mm/kasan/report.c | 11 +++ > >>> 1 file changed, 11 insertions(+) > >>> > >>> diff --git a/mm/kasan/report.c b/mm/kasan/report.c > >>> index c0fb21797550..2485b585004d 100644 > >>> --- a/mm/kasan/report.c > >>> +++ b/mm/kasan/report.c > >>> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > >>> size_t size, bool is_write, > >>> end_report(); > >>> } > >>> > >>> +/** > >>> + * kasan_report - report kasan fault details > >>> + * @addr: valid address of the allocation where the tag fault was > >>> detected > >>> + * @size: size of the allocation where the tag fault was detected > >>> + * @is_write: the instruction that caused the fault was a read or write? > >>> + * @ip: pointer to the instruction that cause the fault > >>> + * > >>> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > >>> + * the address to access the tags, hence it must be valid at this point > >>> in > >>> + * order to not cause a kernel panic. > >>> + */ > >> > >> It doesn't dereference the address, it just checks the tags, right? > >> > >> Ideally, kasan_report() should survive that with HW_TAGS like with the > >> other modes. The reason it doesn't is probably because of a blank > >> addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > >> guess we should somehow check that the memory comes from page_alloc or > >> kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > >> instruction to check whether the memory has tags? > > > > There isn't an architected way to probe whether a memory location has a > > VA->PA mapping. The tags are addressed by PA but you can't reach them if > > you get a page fault on the VA. So we either document the kasan_report() > > preconditions or, as you suggest, update addr_has_metadata() for the > > HW_TAGS case. Something like: > > > > return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > > > > This seems not working on arm64 because according to virt_addr_valid 0 is a > valid virtual address, in fact: > > __is_lm_address(0) == true && pfn_valid(virt_to_pfn(0)) == true. Ah, so __is_lm_address(0) is true. Maybe we should improve this since virt_to_pfn(0) doesn't make much sense. > An option could be to make an exception for virtual address 0 in > addr_has_metadata() something like: > > static inline bool addr_has_metadata(const void *addr) > { > if ((u64)addr == 0) > return false; > > return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); > } As Andrey replied, passing a non-zero small value would still be incorrectly detected as valid. -- Catalin
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 9:32 PM Vincenzo Frascino wrote: > > This seems not working on arm64 because according to virt_addr_valid 0 is a > valid virtual address, in fact: > > __is_lm_address(0) == true && pfn_valid(virt_to_pfn(0)) == true. > > An option could be to make an exception for virtual address 0 in > addr_has_metadata() something like: > > static inline bool addr_has_metadata(const void *addr) > { > if ((u64)addr == 0) > return false; This sounds good to me, but we need to check for < PAGE_SIZE or something like that, right? There's some limit below which accesses are considered null-ptr-derefs. > return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); Do we need is_vmalloc_addr()? As we don't yet have vmalloc support for HW_TAGS.
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: > On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > wrote: > > > > With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > > the address passed as a parameter. > > > > Add a comment to make sure that the preconditions to the function are > > explicitly clarified. > > > > Note: An invalid address (e.g. NULL pointer address) passed to the > > function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > > > > Cc: Andrey Ryabinin > > Cc: Alexander Potapenko > > Cc: Dmitry Vyukov > > Cc: Leon Romanovsky > > Cc: Andrey Konovalov > > Signed-off-by: Vincenzo Frascino > > --- > > mm/kasan/report.c | 11 +++ > > 1 file changed, 11 insertions(+) > > > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c > > index c0fb21797550..2485b585004d 100644 > > --- a/mm/kasan/report.c > > +++ b/mm/kasan/report.c > > @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, size_t > > size, bool is_write, > > end_report(); > > } > > > > +/** > > + * kasan_report - report kasan fault details > > + * @addr: valid address of the allocation where the tag fault was detected > > + * @size: size of the allocation where the tag fault was detected > > + * @is_write: the instruction that caused the fault was a read or write? > > + * @ip: pointer to the instruction that cause the fault > > + * > > + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > > + * the address to access the tags, hence it must be valid at this point in > > + * order to not cause a kernel panic. > > + */ > > It doesn't dereference the address, it just checks the tags, right? > > Ideally, kasan_report() should survive that with HW_TAGS like with the > other modes. The reason it doesn't is probably because of a blank > addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > guess we should somehow check that the memory comes from page_alloc or > kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > instruction to check whether the memory has tags? There isn't an architected way to probe whether a memory location has a VA->PA mapping. The tags are addressed by PA but you can't reach them if you get a page fault on the VA. So we either document the kasan_report() preconditions or, as you suggest, update addr_has_metadata() for the HW_TAGS case. Something like: return is_vmalloc_addr(virt) || virt_addr_valid(virt)); -- Catalin
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino wrote: > > With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > the address passed as a parameter. > > Add a comment to make sure that the preconditions to the function are > explicitly clarified. > > Note: An invalid address (e.g. NULL pointer address) passed to the > function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > > Cc: Andrey Ryabinin > Cc: Alexander Potapenko > Cc: Dmitry Vyukov > Cc: Leon Romanovsky > Cc: Andrey Konovalov > Signed-off-by: Vincenzo Frascino > --- > mm/kasan/report.c | 11 +++ > 1 file changed, 11 insertions(+) > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c > index c0fb21797550..2485b585004d 100644 > --- a/mm/kasan/report.c > +++ b/mm/kasan/report.c > @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, size_t > size, bool is_write, > end_report(); > } > > +/** > + * kasan_report - report kasan fault details > + * @addr: valid address of the allocation where the tag fault was detected > + * @size: size of the allocation where the tag fault was detected > + * @is_write: the instruction that caused the fault was a read or write? > + * @ip: pointer to the instruction that cause the fault > + * > + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > + * the address to access the tags, hence it must be valid at this point in > + * order to not cause a kernel panic. > + */ It doesn't dereference the address, it just checks the tags, right? Ideally, kasan_report() should survive that with HW_TAGS like with the other modes. The reason it doesn't is probably because of a blank addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I guess we should somehow check that the memory comes from page_alloc or kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm instruction to check whether the memory has tags?
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On 1/19/21 6:52 PM, Catalin Marinas wrote: > On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: >> On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino >> wrote: >>> >>> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences >>> the address passed as a parameter. >>> >>> Add a comment to make sure that the preconditions to the function are >>> explicitly clarified. >>> >>> Note: An invalid address (e.g. NULL pointer address) passed to the >>> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. >>> >>> Cc: Andrey Ryabinin >>> Cc: Alexander Potapenko >>> Cc: Dmitry Vyukov >>> Cc: Leon Romanovsky >>> Cc: Andrey Konovalov >>> Signed-off-by: Vincenzo Frascino >>> --- >>> mm/kasan/report.c | 11 +++ >>> 1 file changed, 11 insertions(+) >>> >>> diff --git a/mm/kasan/report.c b/mm/kasan/report.c >>> index c0fb21797550..2485b585004d 100644 >>> --- a/mm/kasan/report.c >>> +++ b/mm/kasan/report.c >>> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, size_t >>> size, bool is_write, >>> end_report(); >>> } >>> >>> +/** >>> + * kasan_report - report kasan fault details >>> + * @addr: valid address of the allocation where the tag fault was detected >>> + * @size: size of the allocation where the tag fault was detected >>> + * @is_write: the instruction that caused the fault was a read or write? >>> + * @ip: pointer to the instruction that cause the fault >>> + * >>> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences >>> + * the address to access the tags, hence it must be valid at this point in >>> + * order to not cause a kernel panic. >>> + */ >> >> It doesn't dereference the address, it just checks the tags, right? >> >> Ideally, kasan_report() should survive that with HW_TAGS like with the >> other modes. The reason it doesn't is probably because of a blank >> addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I >> guess we should somehow check that the memory comes from page_alloc or >> kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm >> instruction to check whether the memory has tags? > > There isn't an architected way to probe whether a memory location has a > VA->PA mapping. The tags are addressed by PA but you can't reach them if > you get a page fault on the VA. So we either document the kasan_report() > preconditions or, as you suggest, update addr_has_metadata() for the > HW_TAGS case. Something like: > > return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > This seems not working on arm64 because according to virt_addr_valid 0 is a valid virtual address, in fact: __is_lm_address(0) == true && pfn_valid(virt_to_pfn(0)) == true. An option could be to make an exception for virtual address 0 in addr_has_metadata() something like: static inline bool addr_has_metadata(const void *addr) { if ((u64)addr == 0) return false; return (is_vmalloc_addr(addr) || virt_addr_valid(addr)); } Let me know what do you think. Thanks! -- Regards, Vincenzo
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 7:57 PM Vincenzo Frascino wrote: > > > > On 1/19/21 6:52 PM, Catalin Marinas wrote: > > On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: > >> On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > >> wrote: > >>> > >>> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > >>> the address passed as a parameter. > >>> > >>> Add a comment to make sure that the preconditions to the function are > >>> explicitly clarified. > >>> > >>> Note: An invalid address (e.g. NULL pointer address) passed to the > >>> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > >>> > >>> Cc: Andrey Ryabinin > >>> Cc: Alexander Potapenko > >>> Cc: Dmitry Vyukov > >>> Cc: Leon Romanovsky > >>> Cc: Andrey Konovalov > >>> Signed-off-by: Vincenzo Frascino > >>> --- > >>> mm/kasan/report.c | 11 +++ > >>> 1 file changed, 11 insertions(+) > >>> > >>> diff --git a/mm/kasan/report.c b/mm/kasan/report.c > >>> index c0fb21797550..2485b585004d 100644 > >>> --- a/mm/kasan/report.c > >>> +++ b/mm/kasan/report.c > >>> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > >>> size_t size, bool is_write, > >>> end_report(); > >>> } > >>> > >>> +/** > >>> + * kasan_report - report kasan fault details > >>> + * @addr: valid address of the allocation where the tag fault was > >>> detected > >>> + * @size: size of the allocation where the tag fault was detected > >>> + * @is_write: the instruction that caused the fault was a read or write? > >>> + * @ip: pointer to the instruction that cause the fault > >>> + * > >>> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > >>> + * the address to access the tags, hence it must be valid at this point > >>> in > >>> + * order to not cause a kernel panic. > >>> + */ > >> > >> It doesn't dereference the address, it just checks the tags, right? > >> > >> Ideally, kasan_report() should survive that with HW_TAGS like with the > >> other modes. The reason it doesn't is probably because of a blank > >> addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > >> guess we should somehow check that the memory comes from page_alloc or > >> kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > >> instruction to check whether the memory has tags? > > > > There isn't an architected way to probe whether a memory location has a > > VA->PA mapping. The tags are addressed by PA but you can't reach them if > > you get a page fault on the VA. So we either document the kasan_report() > > preconditions or, as you suggest, update addr_has_metadata() for the > > HW_TAGS case. Something like: > > > > return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > > > > Or we could have both ;) Sounds good! Please also update the comment to avoid mentioning tag faults. kasan_report() is used for the generic KASAN mode as well, and it doesn't use tags. Thanks!
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On 1/19/21 6:52 PM, Catalin Marinas wrote: > On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: >> On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino >> wrote: >>> >>> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences >>> the address passed as a parameter. >>> >>> Add a comment to make sure that the preconditions to the function are >>> explicitly clarified. >>> >>> Note: An invalid address (e.g. NULL pointer address) passed to the >>> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. >>> >>> Cc: Andrey Ryabinin >>> Cc: Alexander Potapenko >>> Cc: Dmitry Vyukov >>> Cc: Leon Romanovsky >>> Cc: Andrey Konovalov >>> Signed-off-by: Vincenzo Frascino >>> --- >>> mm/kasan/report.c | 11 +++ >>> 1 file changed, 11 insertions(+) >>> >>> diff --git a/mm/kasan/report.c b/mm/kasan/report.c >>> index c0fb21797550..2485b585004d 100644 >>> --- a/mm/kasan/report.c >>> +++ b/mm/kasan/report.c >>> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, size_t >>> size, bool is_write, >>> end_report(); >>> } >>> >>> +/** >>> + * kasan_report - report kasan fault details >>> + * @addr: valid address of the allocation where the tag fault was detected >>> + * @size: size of the allocation where the tag fault was detected >>> + * @is_write: the instruction that caused the fault was a read or write? >>> + * @ip: pointer to the instruction that cause the fault >>> + * >>> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences >>> + * the address to access the tags, hence it must be valid at this point in >>> + * order to not cause a kernel panic. >>> + */ >> >> It doesn't dereference the address, it just checks the tags, right? >> >> Ideally, kasan_report() should survive that with HW_TAGS like with the >> other modes. The reason it doesn't is probably because of a blank >> addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I >> guess we should somehow check that the memory comes from page_alloc or >> kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm >> instruction to check whether the memory has tags? > > There isn't an architected way to probe whether a memory location has a > VA->PA mapping. The tags are addressed by PA but you can't reach them if > you get a page fault on the VA. So we either document the kasan_report() > preconditions or, as you suggest, update addr_has_metadata() for the > HW_TAGS case. Something like: > > return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > Or we could have both ;) -- Regards, Vincenzo
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On 1/19/21 7:02 PM, Catalin Marinas wrote: > On Tue, Jan 19, 2021 at 07:00:57PM +, Vincenzo Frascino wrote: >> On 1/19/21 6:52 PM, Catalin Marinas wrote: >>> On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino wrote: > > With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > the address passed as a parameter. > > Add a comment to make sure that the preconditions to the function are > explicitly clarified. > > Note: An invalid address (e.g. NULL pointer address) passed to the > function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > > Cc: Andrey Ryabinin > Cc: Alexander Potapenko > Cc: Dmitry Vyukov > Cc: Leon Romanovsky > Cc: Andrey Konovalov > Signed-off-by: Vincenzo Frascino > --- > mm/kasan/report.c | 11 +++ > 1 file changed, 11 insertions(+) > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c > index c0fb21797550..2485b585004d 100644 > --- a/mm/kasan/report.c > +++ b/mm/kasan/report.c > @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > size_t size, bool is_write, > end_report(); > } > > +/** > + * kasan_report - report kasan fault details > + * @addr: valid address of the allocation where the tag fault was > detected > + * @size: size of the allocation where the tag fault was detected > + * @is_write: the instruction that caused the fault was a read or write? > + * @ip: pointer to the instruction that cause the fault > + * > + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > + * the address to access the tags, hence it must be valid at this point > in > + * order to not cause a kernel panic. > + */ It doesn't dereference the address, it just checks the tags, right? Ideally, kasan_report() should survive that with HW_TAGS like with the other modes. The reason it doesn't is probably because of a blank addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I guess we should somehow check that the memory comes from page_alloc or kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm instruction to check whether the memory has tags? >>> >>> There isn't an architected way to probe whether a memory location has a >>> VA->PA mapping. The tags are addressed by PA but you can't reach them if >>> you get a page fault on the VA. So we either document the kasan_report() >>> preconditions or, as you suggest, update addr_has_metadata() for the >>> HW_TAGS case. Something like: >>> >>> return is_vmalloc_addr(virt) || virt_addr_valid(virt)); >> >> Or we could have both ;) > > True. Documentation doesn't hurt (well, only when it's wrong ;)). > Testing the patch now, I will send it in half an hour. -- Regards, Vincenzo
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
On Tue, Jan 19, 2021 at 07:00:57PM +, Vincenzo Frascino wrote: > On 1/19/21 6:52 PM, Catalin Marinas wrote: > > On Tue, Jan 19, 2021 at 07:27:43PM +0100, Andrey Konovalov wrote: > >> On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > >> wrote: > >>> > >>> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences > >>> the address passed as a parameter. > >>> > >>> Add a comment to make sure that the preconditions to the function are > >>> explicitly clarified. > >>> > >>> Note: An invalid address (e.g. NULL pointer address) passed to the > >>> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. > >>> > >>> Cc: Andrey Ryabinin > >>> Cc: Alexander Potapenko > >>> Cc: Dmitry Vyukov > >>> Cc: Leon Romanovsky > >>> Cc: Andrey Konovalov > >>> Signed-off-by: Vincenzo Frascino > >>> --- > >>> mm/kasan/report.c | 11 +++ > >>> 1 file changed, 11 insertions(+) > >>> > >>> diff --git a/mm/kasan/report.c b/mm/kasan/report.c > >>> index c0fb21797550..2485b585004d 100644 > >>> --- a/mm/kasan/report.c > >>> +++ b/mm/kasan/report.c > >>> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, > >>> size_t size, bool is_write, > >>> end_report(); > >>> } > >>> > >>> +/** > >>> + * kasan_report - report kasan fault details > >>> + * @addr: valid address of the allocation where the tag fault was > >>> detected > >>> + * @size: size of the allocation where the tag fault was detected > >>> + * @is_write: the instruction that caused the fault was a read or write? > >>> + * @ip: pointer to the instruction that cause the fault > >>> + * > >>> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences > >>> + * the address to access the tags, hence it must be valid at this point > >>> in > >>> + * order to not cause a kernel panic. > >>> + */ > >> > >> It doesn't dereference the address, it just checks the tags, right? > >> > >> Ideally, kasan_report() should survive that with HW_TAGS like with the > >> other modes. The reason it doesn't is probably because of a blank > >> addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > >> guess we should somehow check that the memory comes from page_alloc or > >> kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > >> instruction to check whether the memory has tags? > > > > There isn't an architected way to probe whether a memory location has a > > VA->PA mapping. The tags are addressed by PA but you can't reach them if > > you get a page fault on the VA. So we either document the kasan_report() > > preconditions or, as you suggest, update addr_has_metadata() for the > > HW_TAGS case. Something like: > > > > return is_vmalloc_addr(virt) || virt_addr_valid(virt)); > > Or we could have both ;) True. Documentation doesn't hurt (well, only when it's wrong ;)). -- Catalin
Re: [PATCH] kasan: Add explicit preconditions to kasan_report()
Hi Andrey, On 1/19/21 6:27 PM, Andrey Konovalov wrote: > On Tue, Jan 19, 2021 at 6:26 PM Vincenzo Frascino > wrote: >> >> With the introduction of KASAN_HW_TAGS, kasan_report() dereferences >> the address passed as a parameter. >> >> Add a comment to make sure that the preconditions to the function are >> explicitly clarified. >> >> Note: An invalid address (e.g. NULL pointer address) passed to the >> function when, KASAN_HW_TAGS is enabled, leads to a kernel panic. >> >> Cc: Andrey Ryabinin >> Cc: Alexander Potapenko >> Cc: Dmitry Vyukov >> Cc: Leon Romanovsky >> Cc: Andrey Konovalov >> Signed-off-by: Vincenzo Frascino >> --- >> mm/kasan/report.c | 11 +++ >> 1 file changed, 11 insertions(+) >> >> diff --git a/mm/kasan/report.c b/mm/kasan/report.c >> index c0fb21797550..2485b585004d 100644 >> --- a/mm/kasan/report.c >> +++ b/mm/kasan/report.c >> @@ -403,6 +403,17 @@ static void __kasan_report(unsigned long addr, size_t >> size, bool is_write, >> end_report(); >> } >> >> +/** >> + * kasan_report - report kasan fault details >> + * @addr: valid address of the allocation where the tag fault was detected >> + * @size: size of the allocation where the tag fault was detected >> + * @is_write: the instruction that caused the fault was a read or write? >> + * @ip: pointer to the instruction that cause the fault >> + * >> + * Note: When CONFIG_KASAN_HW_TAGS is enabled kasan_report() dereferences >> + * the address to access the tags, hence it must be valid at this point in >> + * order to not cause a kernel panic. >> + */ > > It doesn't dereference the address, it just checks the tags, right? > This is correct, just realized that the use of "dereference" here is misleading. > Ideally, kasan_report() should survive that with HW_TAGS like with the > other modes. The reason it doesn't is probably because of a blank > addr_has_metadata() definition for HW_TAGS in mm/kasan/kasan.h. I > guess we should somehow check that the memory comes from page_alloc or > kmalloc. Or otherwise make sure that it has tags. Maybe there's an arm > instruction to check whether the memory has tags? > I agree, looking a second time at the code the problem comes from addr_has_metadata(): ... [ 18.127273] BUG: KASAN: invalid-access in 0x0 [ 18.128604] Read at addr by task swapper/0/1 [ 18.130311] Unable to handle kernel NULL pointer dereference at virtual address [ 18.131291] Mem abort info: [ 18.131696] ESR = 0x9604 [ 18.132169] EC = 0x25: DABT (current EL), IL = 32 bits [ 18.132953] SET = 0, FnV = 0 [ 18.133433] EA = 0, S1PTW = 0 [ 18.133907] Data abort info: [ 18.134308] ISV = 0, ISS = 0x0004 [ 18.134883] CM = 0, WnR = 0 [ 18.135436] [] user address but active_mm is swapper [ 18.136372] Internal error: Oops: 9604 [#1] PREEMPT SMP [ 18.137280] Modules linked in: [ 18.138182] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.11.0-rc4-7-g86cba71f117-dirty #2 [ 18.139275] Hardware name: QEMU QEMU Virtual Machine, BIOS 0.0.0 02/06/2015 [ 18.140342] pstate: 60400085 (nZCv daIf +PAN -UAO -TCO BTYPE=--) [ 18.141324] pc : mte_get_mem_tag+0x24/0x40 [ 18.142487] lr : print_tags+0x1c/0x40 [ 18.143095] sp : 80001004bcf0 [ 18.143570] x29: 80001004bcf0 x28: [ 18.144526] x27: d042f0bf04e0 x26: d042f0ca1068 [ 18.145369] x25: d042f0bdde58 x24: d042f1458000 [ 18.146209] x23: x22: [ 18.147047] x21: x20: [ 18.147928] x19: x18: [ 18.148928] x17: 000e x16: 0001 [ 18.149837] x15: 80009004ba17 x14: 0006 [ 18.150774] x13: d042f11b27e0 x12: 0399 [ 18.151653] x11: 0133 x10: d042f11b27e0 [ 18.152544] x9 : d042f11b27e0 x8 : efff [ 18.153443] x7 : d042f120a7e0 x6 : d042f120a7e0 [ 18.154272] x5 : bff4 x4 : [ 18.155096] x3 : x2 : [ 18.155958] x1 : x0 : [ 18.157145] Call trace: [ 18.157615] mte_get_mem_tag+0x24/0x40 [ 18.158258] kasan_report+0xec/0x1b0 ... I noticed it differently but you can easily reproduce it calling kasan_report(0,0,0,0); from somewhere. I will send a patch tomorrow that checks if the memory comes from page_alloc or kmalloc. Not sure what you mean for "instruction to check whether the memory has tags". Thanks! -- Regards, Vincenzo