similar to: [PATCH AUTOSEL for 4.4 34/36] x86/mm/kmmio: Fix mmiotrace for page unaligned addresses

If something calls ioremap with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but the entire page was armed for mmiotracing. On calling iounmap the address passed to unregister_kmmio_probe was PAGE_SIZE aligned by the caller leading to a complete freeze of the machine. We should always page

From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but the entire page was armed for mmiotracing. On calling iounmap() the address passed to

From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but the entire page was armed for mmiotracing. On calling iounmap() the address passed to

From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but the entire page was armed for mmiotracing. On calling iounmap() the address passed to

4.14-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but

3.18-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but

4.4-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but

4.9-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <kherbst at redhat.com> [ Upstream commit 6d60ce384d1d5ca32b595244db4077a419acc687 ] If something calls ioremap() with an address not aligned to PAGE_SIZE, the returned address might be not aligned as well. This led to a probe registered on exactly the returned address, but

From: Karol Herbst <nouveau at karolherbst.de> This patch has been added to the 4.1 stable tree. If you have any objections, please let us know. =============== [ Upstream commit cfa52c0cfa4d727aa3e457bf29aeff296c528a08 ] Because Linux might use bigger pages than the 4K pages to handle those mmio ioremaps, the kmmio code shouldn't rely on the pade id as it currently does. Using the

From: Karol Herbst <nouveau at karolherbst.de> This patch has been added to the 3.18 stable tree. If you have any objections, please let us know. =============== [ Upstream commit cfa52c0cfa4d727aa3e457bf29aeff296c528a08 ] Because Linux might use bigger pages than the 4K pages to handle those mmio ioremaps, the kmmio code shouldn't rely on the pade id as it currently does. Using the

Because Linux might use bigger pages than the 4K pages to handle those mmio ioremaps, the kmmio code shouldn't rely on the pade id as it currently does. Using the memory address instead of the page id let's us lookup how big the page is and what it's base address is, so that we won't get a page fault within the same page twice anymore. I don't know if I got this right though,

4.5-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <nouveau at karolherbst.de> commit cfa52c0cfa4d727aa3e457bf29aeff296c528a08 upstream. Because Linux might use bigger pages than the 4K pages to handle those mmio ioremaps, the kmmio code shouldn't rely on the pade id as it currently does. Using the memory address instead

4.4-stable review patch. If anyone has any objections, please let me know. ------------------ From: Karol Herbst <nouveau at karolherbst.de> commit cfa52c0cfa4d727aa3e457bf29aeff296c528a08 upstream. Because Linux might use bigger pages than the 4K pages to handle those mmio ioremaps, the kmmio code shouldn't rely on the pade id as it currently does. Using the memory address instead

After every iounmap mmiotrace has to free kmmio_fault_pages, but it can't do it directly, so it defers freeing by RCU. It usually works, but when mmiotraced code calls ioremap-iounmap multiple times without sleeping between (so RCU won't kick in and start freeing) it can be given the same virtual address, so at every iounmap mmiotrace will schedule the same pages for release. Obviously it

After every iounmap mmiotrace has to free kmmio_fault_pages, but it can't do it directly, so it defers freeing by RCU. It usually works, but when mmiotraced code calls ioremap-iounmap multiple times without sleeping between (so RCU won't kick in and start freeing) it can be given the same virtual address, so at every iounmap mmiotrace will schedule the same pages for release. Obviously it

Hi, I think I got mmiotracing back to work on my machine. Would you please test this patch on your machines too? I still got some UNKNOWN entries in the mmiotrace file, but besides that it seems to work. I have no idea if that's the right way though, and it needs a big cleanup, but I just wanted to see if the basic idea works the patch should apply just fine on any 4.4 kernel happy

Hi Linus, The following changes since commit 540ad6b62b3a188a53b51cac81d8a60d40e29fbd: Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/linux-2.6-kgdb (2010-07-28 20:01:26 -0700) are available in the git repository at: git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-mm-for-linus Andres Salomon (1): x86, mm: Create

add_kmmio_fault_page using -1 instead of the -ENOMEM defined macro to specify kmmio_fault_page allocation failed. Smatch tool warning: arch/x86/mm/kmmio.c:389 add_kmmio_fault_page() warn: returning -1 instead of -ENOMEM is sloppy Signed-off-by: Vasyl Gomonovych <gomonovych at gmail.com> --- arch/x86/mm/kmmio.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git

Replacing return value -1 to error code Signed-off-by: Mugilraj Dhavachelvan <dmugil2000 at gmail.com> --- arch/x86/mm/kmmio.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/mm/kmmio.c b/arch/x86/mm/kmmio.c index be020a7bc414..15430520c232 100644 --- a/arch/x86/mm/kmmio.c +++ b/arch/x86/mm/kmmio.c @@ -386,7 +386,7 @@ static int add_kmmio_fault_page(unsigned

Both call sites, that access kmmio_ctx, access kmmio_ctx with interrupts disabled. There is no need to use get_cpu_var() which additionally disables preemption. Use this_cpu_ptr() to access the kmmio_ctx variable of the current CPU. Signed-off-by: Sebastian Andrzej Siewior <bigeasy at linutronix.de> --- arch/x86/mm/kmmio.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-)