| Message ID | 20240306232339.29659-1-richard@nod.at (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [1/2,RFC] proc: pagemap: Expose whether a PTE is writable | expand |
On 3/7/24 4:23 AM, Richard Weinberger wrote: > Is a PTE present and writable, bit 58 will be set. > This allows detecting CoW memory mappings and other mappings > where a write access will cause a page fault. > > Signed-off-by: Richard Weinberger <richard@nod.at> > --- > fs/proc/task_mmu.c | 8 ++++++++ > 1 file changed, 8 insertions(+) > > diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c > index 3f78ebbb795f..7c7e0e954c02 100644 > --- a/fs/proc/task_mmu.c > +++ b/fs/proc/task_mmu.c > @@ -1341,6 +1341,7 @@ struct pagemapread { > #define PM_SOFT_DIRTY BIT_ULL(55) > #define PM_MMAP_EXCLUSIVE BIT_ULL(56) > #define PM_UFFD_WP BIT_ULL(57) > +#define PM_WRITE BIT_ULL(58) The name doesn't mention present from its "present and writable" definition. Maybe some other name like PM_PRESENT_WRITE? > #define PM_FILE BIT_ULL(61) > #define PM_SWAP BIT_ULL(62) > #define PM_PRESENT BIT_ULL(63) > @@ -1417,6 +1418,8 @@ static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm, > flags |= PM_SOFT_DIRTY; > if (pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > + if (pte_write(pte)) > + flags |= PM_WRITE; > } else if (is_swap_pte(pte)) { > swp_entry_t entry; > if (pte_swp_soft_dirty(pte)) > @@ -1483,6 +1486,8 @@ static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end, > flags |= PM_SOFT_DIRTY; > if (pmd_uffd_wp(pmd)) > flags |= PM_UFFD_WP; > + if (pmd_write(pmd)) > + flags |= PM_WRITE; > if (pm->show_pfn) > frame = pmd_pfn(pmd) + > ((addr & ~PMD_MASK) >> PAGE_SHIFT); > @@ -1586,6 +1591,9 @@ static int pagemap_hugetlb_range(pte_t *ptep, unsigned long hmask, > if (huge_pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > > + if (pte_write(pte)) > + flags |= PM_WRITE; > + > flags |= PM_PRESENT; > if (pm->show_pfn) > frame = pte_pfn(pte) +
On 07.03.24 00:23, Richard Weinberger wrote: > Is a PTE present and writable, bit 58 will be set. > This allows detecting CoW memory mappings and other mappings > where a write access will cause a page fault. > But why is that required? What is the target use case? (I did not get the cover letter in my inbox) We're running slowly but steadily out of bits, so we better make wise decisions. Also, consider: Architectures where the dirty/access bit is not HW managed could indicate "writable" here although we *will* get a page fault to set the page dirty/accessed. So best this can universally do is say "this PTE currently has write permissions". > Signed-off-by: Richard Weinberger <richard@nod.at> > --- > fs/proc/task_mmu.c | 8 ++++++++ > 1 file changed, 8 insertions(+) > > diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c > index 3f78ebbb795f..7c7e0e954c02 100644 > --- a/fs/proc/task_mmu.c > +++ b/fs/proc/task_mmu.c > @@ -1341,6 +1341,7 @@ struct pagemapread { > #define PM_SOFT_DIRTY BIT_ULL(55) > #define PM_MMAP_EXCLUSIVE BIT_ULL(56) > #define PM_UFFD_WP BIT_ULL(57) > +#define PM_WRITE BIT_ULL(58) > #define PM_FILE BIT_ULL(61) > #define PM_SWAP BIT_ULL(62) > #define PM_PRESENT BIT_ULL(63) > @@ -1417,6 +1418,8 @@ static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm, > flags |= PM_SOFT_DIRTY; > if (pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > + if (pte_write(pte)) > + flags |= PM_WRITE; > } else if (is_swap_pte(pte)) { > swp_entry_t entry; > if (pte_swp_soft_dirty(pte)) > @@ -1483,6 +1486,8 @@ static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end, > flags |= PM_SOFT_DIRTY; > if (pmd_uffd_wp(pmd)) > flags |= PM_UFFD_WP; > + if (pmd_write(pmd)) > + flags |= PM_WRITE; > if (pm->show_pfn) > frame = pmd_pfn(pmd) + > ((addr & ~PMD_MASK) >> PAGE_SHIFT); > @@ -1586,6 +1591,9 @@ static int pagemap_hugetlb_range(pte_t *ptep, unsigned long hmask, > if (huge_pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > > + if (pte_write(pte)) > + flags |= PM_WRITE; > + > flags |= PM_PRESENT; > if (pm->show_pfn) > frame = pte_pfn(pte) +
----- Ursprüngliche Mail ----- > Von: "David Hildenbrand" <david@redhat.com> > But why is that required? What is the target use case? (I did not get > the cover letter in my inbox) > > We're running slowly but steadily out of bits, so we better make wise > decisions. > > Also, consider: Architectures where the dirty/access bit is not HW > managed could indicate "writable" here although we *will* get a page > fault to set the page dirty/accessed. I'm currently investigating why a real-time application faces unexpected page faults. Page faults are usually fatal for real-time work loads because the latency constraints are no longer met. So, I wrote a small tool to inspect the memory mappings of a process to find areas which are not correctly pre-faulted. While doing so I noticed that there is currently no way to detect CoW mappings. Exposing the writable property of a PTE seemed like a good start to me. > So best this can universally do is say "this PTE currently has write > permissions". Ok. Thanks, //richard
On 07.03.24 12:10, Richard Weinberger wrote: > ----- Ursprüngliche Mail ----- >> Von: "David Hildenbrand" <david@redhat.com> >> But why is that required? What is the target use case? (I did not get >> the cover letter in my inbox) >> >> We're running slowly but steadily out of bits, so we better make wise >> decisions. >> >> Also, consider: Architectures where the dirty/access bit is not HW >> managed could indicate "writable" here although we *will* get a page >> fault to set the page dirty/accessed. > > I'm currently investigating why a real-time application faces unexpected > page faults. Page faults are usually fatal for real-time work loads because > the latency constraints are no longer met. Are you concerned about any type of page fault, or are things like a simple remapping of the same page from "read-only to writable" acceptable? ("very minor fault") > > So, I wrote a small tool to inspect the memory mappings of a process to find > areas which are not correctly pre-faulted. While doing so I noticed that > there is currently no way to detect CoW mappings. > Exposing the writable property of a PTE seemed like a good start to me. Is it just about "detection" for debugging purposes or about "fixup" in running applications? If it's the latter, MADV_POPULATE_WRITE might do what you want (in writable mappings).
----- Ursprüngliche Mail ----- > Von: "David Hildenbrand" <david@redhat.com> >> I'm currently investigating why a real-time application faces unexpected >> page faults. Page faults are usually fatal for real-time work loads because >> the latency constraints are no longer met. > > Are you concerned about any type of page fault, or are things like a > simple remapping of the same page from "read-only to writable" > acceptable? ("very minor fault") Any page fault has to be avoided. To give you more background, the real time application runs on Xenomai, a real time extension for Linux. Xenomai applies already many tweaks to the kernel to trigger pre-faulting of memory areas. But sometimes the application does not use the Xenomai API correctly or there is an bug in Xenomai it self. Currently I'm suspecting the latter. >> >> So, I wrote a small tool to inspect the memory mappings of a process to find >> areas which are not correctly pre-faulted. While doing so I noticed that >> there is currently no way to detect CoW mappings. >> Exposing the writable property of a PTE seemed like a good start to me. > > Is it just about "detection" for debugging purposes or about "fixup" in > running applications? It's only about debugging. If an application fails a test I want to have a tool which tells me what memory mappings are wonky or could cause a fault at runtime. I fully understand that my use case is a corner case and anything but mainline. While developing my debug tool I thought that improving the pagemap interface might help others too. Thanks, //richard
On 07.03.24 12:51, Richard Weinberger wrote: > ----- Ursprüngliche Mail ----- >> Von: "David Hildenbrand" <david@redhat.com> >>> I'm currently investigating why a real-time application faces unexpected >>> page faults. Page faults are usually fatal for real-time work loads because >>> the latency constraints are no longer met. >> >> Are you concerned about any type of page fault, or are things like a >> simple remapping of the same page from "read-only to writable" >> acceptable? ("very minor fault") > > Any page fault has to be avoided. > To give you more background, the real time application runs on Xenomai, > a real time extension for Linux. > Xenomai applies already many tweaks to the kernel to trigger pre-faulting of > memory areas. But sometimes the application does not use the Xenomai API > correctly or there is an bug in Xenomai it self. > Currently I'm suspecting the latter. > Thanks for the details! >>> >>> So, I wrote a small tool to inspect the memory mappings of a process to find >>> areas which are not correctly pre-faulted. While doing so I noticed that >>> there is currently no way to detect CoW mappings. >>> Exposing the writable property of a PTE seemed like a good start to me. >> >> Is it just about "detection" for debugging purposes or about "fixup" in >> running applications? > > It's only about debugging. If an application fails a test I want to have > a tool which tells me what memory mappings are wonky or could cause a fault > at runtime. One destructive way to find out in a writable mapping if the page would actually get remapped: a) Read the PFN of a virtual address using pagemap b) Write to the virtual address using /proc/pid/mem c) Read the PFN of a virtual address using pagemap to see if it changed If the application can be paused, you could read+write a single byte, turning it non-destructive. But that would still "hide" the remap-writable-type faults. > > I fully understand that my use case is a corner case and anything but mainline. > While developing my debug tool I thought that improving the pagemap interface > might help others too. I'm fine with this (can be a helpful debugging tool for some other cases as well, and IIRC we don't have another interface to introspect this), as long as we properly document the corner case that there could still be writefaults on some architectures when the page would not be accessed/dirty yet.
On 07.03.24 12:59, David Hildenbrand wrote: > On 07.03.24 12:51, Richard Weinberger wrote: >> ----- Ursprüngliche Mail ----- >>> Von: "David Hildenbrand" <david@redhat.com> >>>> I'm currently investigating why a real-time application faces unexpected >>>> page faults. Page faults are usually fatal for real-time work loads because >>>> the latency constraints are no longer met. >>> >>> Are you concerned about any type of page fault, or are things like a >>> simple remapping of the same page from "read-only to writable" >>> acceptable? ("very minor fault") >> >> Any page fault has to be avoided. >> To give you more background, the real time application runs on Xenomai, >> a real time extension for Linux. >> Xenomai applies already many tweaks to the kernel to trigger pre-faulting of >> memory areas. But sometimes the application does not use the Xenomai API >> correctly or there is an bug in Xenomai it self. >> Currently I'm suspecting the latter. >> > > Thanks for the details! > >>>> >>>> So, I wrote a small tool to inspect the memory mappings of a process to find >>>> areas which are not correctly pre-faulted. While doing so I noticed that >>>> there is currently no way to detect CoW mappings. >>>> Exposing the writable property of a PTE seemed like a good start to me. >>> >>> Is it just about "detection" for debugging purposes or about "fixup" in >>> running applications? >> >> It's only about debugging. If an application fails a test I want to have >> a tool which tells me what memory mappings are wonky or could cause a fault >> at runtime. > > One destructive way to find out in a writable mapping if the page would > actually get remapped: > > a) Read the PFN of a virtual address using pagemap > b) Write to the virtual address using /proc/pid/mem > c) Read the PFN of a virtual address using pagemap to see if it changed > > If the application can be paused, you could read+write a single byte, > turning it non-destructive. > > But that would still "hide" the remap-writable-type faults. > >> >> I fully understand that my use case is a corner case and anything but mainline. >> While developing my debug tool I thought that improving the pagemap interface >> might help others too. > > I'm fine with this (can be a helpful debugging tool for some other cases > as well, and IIRC we don't have another interface to introspect this), > as long as we properly document the corner case that there could still > be writefaults on some architectures when the page would not be > accessed/dirty yet. > [and I just recall, there are some other corner cases. For example, pages in a shadow stack can be pte_write(), but they can only be written by HW indirectly when modifying the stack, and ordinary write access would still fault]
----- Ursprüngliche Mail ----- > Von: "David Hildenbrand" <david@redhat.com> >> One destructive way to find out in a writable mapping if the page would >> actually get remapped: >> >> a) Read the PFN of a virtual address using pagemap >> b) Write to the virtual address using /proc/pid/mem >> c) Read the PFN of a virtual address using pagemap to see if it changed >> >> If the application can be paused, you could read+write a single byte, >> turning it non-destructive. I'm not so sure whether this works well if a mapping is device memory or such. >> But that would still "hide" the remap-writable-type faults. Xenomai will tell me anyway when there was a page fault while a real time thread had the CPU. My idea was having a tool to check before the applications enters the critical phase. >>> I fully understand that my use case is a corner case and anything but mainline. >>> While developing my debug tool I thought that improving the pagemap interface >>> might help others too. >> >> I'm fine with this (can be a helpful debugging tool for some other cases >> as well, and IIRC we don't have another interface to introspect this), >> as long as we properly document the corner case that there could still >> be writefaults on some architectures when the page would not be >> accessed/dirty yet. Cool. :) > > [and I just recall, there are some other corner cases. For example, > pages in a shadow stack can be pte_write(), but they can only be written > by HW indirectly when modifying the stack, and ordinary write access > would still fault] Yeah, I noticed this while browsing through various pte_write() implementations. That's a tradeoff I can live with. Thanks, //richard
On Thu, Mar 07, 2024 at 12:23:38AM +0100, Richard Weinberger wrote: > Is a PTE present and writable, bit 58 will be set. > This allows detecting CoW memory mappings and other mappings > where a write access will cause a page fault. I think David has highlighted it elsewhere in the thread, but this explanation definitely needs bulking up. Need to emphsaise that we detect cases where a fault will occur (_possibly_ CoW, _possibly_ write notify clean file-backed page, _possibly_ other cases where we need write fault tracking). Very important to differentiate between a _page table_ read/write flag being set and the mapping being read-only, it's a concern that being loose on this might confuse people somewhat. > > Signed-off-by: Richard Weinberger <richard@nod.at> > --- > fs/proc/task_mmu.c | 8 ++++++++ > 1 file changed, 8 insertions(+) > > diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c > index 3f78ebbb795f..7c7e0e954c02 100644 > --- a/fs/proc/task_mmu.c > +++ b/fs/proc/task_mmu.c > @@ -1341,6 +1341,7 @@ struct pagemapread { > #define PM_SOFT_DIRTY BIT_ULL(55) > #define PM_MMAP_EXCLUSIVE BIT_ULL(56) > #define PM_UFFD_WP BIT_ULL(57) > +#define PM_WRITE BIT_ULL(58) As an extension of the above comment re: confusion, I really dislike PM_WRITE. Something like PM_PTE_WRITABLE might be better? > #define PM_FILE BIT_ULL(61) > #define PM_SWAP BIT_ULL(62) > #define PM_PRESENT BIT_ULL(63) > @@ -1417,6 +1418,8 @@ static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm, > flags |= PM_SOFT_DIRTY; > if (pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > + if (pte_write(pte)) > + flags |= PM_WRITE; > } else if (is_swap_pte(pte)) { > swp_entry_t entry; > if (pte_swp_soft_dirty(pte)) > @@ -1483,6 +1486,8 @@ static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end, > flags |= PM_SOFT_DIRTY; > if (pmd_uffd_wp(pmd)) > flags |= PM_UFFD_WP; > + if (pmd_write(pmd)) > + flags |= PM_WRITE; > if (pm->show_pfn) > frame = pmd_pfn(pmd) + > ((addr & ~PMD_MASK) >> PAGE_SHIFT); > @@ -1586,6 +1591,9 @@ static int pagemap_hugetlb_range(pte_t *ptep, unsigned long hmask, > if (huge_pte_uffd_wp(pte)) > flags |= PM_UFFD_WP; > > + if (pte_write(pte)) This should be huge_pte_write(). It amounts to the same thing, but for consistency :) > + flags |= PM_WRITE; > + > flags |= PM_PRESENT; > if (pm->show_pfn) > frame = pte_pfn(pte) + > -- > 2.35.3 > Overall I _really_ like the idea of exposing this. Not long ago I wanted to be able to assess whether private mappings were CoW'd or not 'at a glance' and couldn't find any means of doing this (of course I might have missed something but I don't think there is anything). So I think a single bit in /proc/$pid/pagemap is absolutely worthwhile to get this information. I'd like to see a non-RFC version submitted :) as discussed on irc, probably best after merge window!
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c index 3f78ebbb795f..7c7e0e954c02 100644 --- a/fs/proc/task_mmu.c +++ b/fs/proc/task_mmu.c @@ -1341,6 +1341,7 @@ struct pagemapread { #define PM_SOFT_DIRTY BIT_ULL(55) #define PM_MMAP_EXCLUSIVE BIT_ULL(56) #define PM_UFFD_WP BIT_ULL(57) +#define PM_WRITE BIT_ULL(58) #define PM_FILE BIT_ULL(61) #define PM_SWAP BIT_ULL(62) #define PM_PRESENT BIT_ULL(63) @@ -1417,6 +1418,8 @@ static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm, flags |= PM_SOFT_DIRTY; if (pte_uffd_wp(pte)) flags |= PM_UFFD_WP; + if (pte_write(pte)) + flags |= PM_WRITE; } else if (is_swap_pte(pte)) { swp_entry_t entry; if (pte_swp_soft_dirty(pte)) @@ -1483,6 +1486,8 @@ static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end, flags |= PM_SOFT_DIRTY; if (pmd_uffd_wp(pmd)) flags |= PM_UFFD_WP; + if (pmd_write(pmd)) + flags |= PM_WRITE; if (pm->show_pfn) frame = pmd_pfn(pmd) + ((addr & ~PMD_MASK) >> PAGE_SHIFT); @@ -1586,6 +1591,9 @@ static int pagemap_hugetlb_range(pte_t *ptep, unsigned long hmask, if (huge_pte_uffd_wp(pte)) flags |= PM_UFFD_WP; + if (pte_write(pte)) + flags |= PM_WRITE; + flags |= PM_PRESENT; if (pm->show_pfn) frame = pte_pfn(pte) +
Is a PTE present and writable, bit 58 will be set. This allows detecting CoW memory mappings and other mappings where a write access will cause a page fault. Signed-off-by: Richard Weinberger <richard@nod.at> --- fs/proc/task_mmu.c | 8 ++++++++ 1 file changed, 8 insertions(+)