BMC/Intel-BMC/linux.git/arch/s390/include/asm/page.h, branch dev

revert "s390/mm: make hugepages_supported a boot time decision"

2015-07-17T23:39:52+00:00

Heiko noticed that the current check for hugepage support on s390 is a little bit too harsh as systems which do not support will crash. The reason is that pageblock_order can now get negative when we set HPAGE_SHIFT to 0. To avoid all this and to avoid opening another can of worms with enabling HUGETLB_PAGE_SIZE_VARIABLE I think it would be best to simply allow architectures to define their own hugepages_supported(). Revert bea41197ead3 ("s390/mm: make hugepages_supported a boot time decision") in preparation. Signed-off-by: Dominik Dingel Acked-by: Martin Schwidefsky Cc: Heiko Carstens Cc: Christian Borntraeger Cc: Michael Holzheu Cc: Gerald Schaefer Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

revert "s390/mm: change HPAGE_SHIFT type to int"

2015-07-17T23:39:52+00:00

Heiko noticed that the current check for hugepage support on s390 is a little bit too harsh as systems which do not support will crash. The reason is that pageblock_order can now get negative when we set HPAGE_SHIFT to 0. To avoid all this and to avoid opening another can of worms with enabling HUGETLB_PAGE_SIZE_VARIABLE I think it would be best to simply allow architectures to define their own hugepages_supported(). This patch (of 4): revert commit cf54e2fce51c ("s390/mm: change HPAGE_SHIFT type to int") in preparation. Signed-off-by: Dominik Dingel Acked-by: Martin Schwidefsky Cc: Heiko Carstens Cc: Christian Borntraeger Cc: Michael Holzheu Cc: Gerald Schaefer Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

s390/mm: change HPAGE_SHIFT type to int

2015-06-26T00:00:35+00:00

With making HPAGE_SHIFT an unsigned integer we also accidentally changed pageblock_order. In order to avoid compiler warnings we make HPAGE_SHFIT an int again. Signed-off-by: Dominik Dingel Suggested-by: Andrew Morton Cc: Martin Schwidefsky Cc: Heiko Carstens Cc: Christian Borntraeger Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

s390/mm: make hugepages_supported a boot time decision

2015-06-26T00:00:35+00:00

There is a potential bug with KVM and hugetlbfs if the hardware does not support hugepages (EDAT1). We fix this by making EDAT1 a hard requirement for hugepages and therefore removing and simplifying code. As s390, with the sw-emulated hugepages, was the only user of arch_prepare/release_hugepage I also removed theses calls from common and other architecture code. This patch (of 5): By dropping support for hugepages on machines which do not have the hardware feature EDAT1, we fix a potential s390 KVM bug. The bug would happen if a guest is backed by hugetlbfs (not supported currently), but does not get pagetables with PGSTE. This would lead to random memory overwrites. Signed-off-by: Dominik Dingel Acked-by: Martin Schwidefsky Cc: Heiko Carstens Cc: Christian Borntraeger Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

s390: let the compiler do page clearing

2015-02-26T08:24:49+00:00

The hardware folks told me that for page clearing "when you exactly know what to do, hand written xc+pfd is usally faster then mvcl for page clearing, as it saves millicode overhead and parameter parsing and checking" as long as you dont need the cache bypassing. Turns out that gcc already does a proper xc,pfd loop. A small test on z196 that does buff = mmap(NULL, bufsize,PROT_EXEC|PROT_WRITE|PROT_READ,AP_PRIVATE| MAP_ANONYMOUS,0,0); for ( i = 0; i < bufsize; i+= 256) buff[i] = 0x5; gets 20% faster (touches every cache line of a page) and buff = mmap(NULL, bufsize,PROT_EXEC|PROT_WRITE|PROT_READ,AP_PRIVATE| MAP_ANONYMOUS,0,0); for ( i = 0; i < bufsize; i+= 4096) buff[i] = 0x5; is within noise ratio (touches one cache line of a page). As the clear_page is usually called for first memory accesses we can assume that at least one cache line is used afterwards, so this change should be always better. Another benchmark, a make -j 40 of my testsuite in tmpfs with hot caches on a 32cpu system: -- unpatched -- -- patched -- real 0m1.017s real 0m0.994s (~2% faster, but in noise) user 0m5.339s user 0m5.016s (~6% faster) sys 0m0.691s sys 0m0.632s (~8% faster) Let use the same define to memset as the asm-generic variant Signed-off-by: Christian Borntraeger Signed-off-by: Martin Schwidefsky

arm64,ia64,ppc,s390,sh,tile,um,x86,mm: remove default gate area

2014-08-08T22:57:27+00:00

The core mm code will provide a default gate area based on FIXADDR_USER_START and FIXADDR_USER_END if !defined(__HAVE_ARCH_GATE_AREA) && defined(AT_SYSINFO_EHDR). This default is only useful for ia64. arm64, ppc, s390, sh, tile, 64-bit UML, and x86_32 have their own code just to disable it. arm, 32-bit UML, and x86_64 have gate areas, but they have their own implementations. This gets rid of the default and moves the code into ia64. This should save some code on architectures without a gate area: it's now possible to inline the gate_area functions in the default case. Signed-off-by: Andy Lutomirski Acked-by: Nathan Lynch Acked-by: H. Peter Anvin Acked-by: Benjamin Herrenschmidt [in principle] Acked-by: Richard Weinberger [for um] Acked-by: Will Deacon [for arm64] Cc: Catalin Marinas Cc: Will Deacon Cc: Tony Luck Cc: Fenghua Yu Cc: Benjamin Herrenschmidt Cc: Paul Mackerras Cc: Martin Schwidefsky Cc: Heiko Carstens Cc: Chris Metcalf Cc: Jeff Dike Cc: Richard Weinberger Cc: Thomas Gleixner Cc: Ingo Molnar Cc: "H. Peter Anvin" Cc: Nathan Lynch Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

s390/mm: optimize copy_page

2013-11-20T08:04:55+00:00

Always use the mvcl instruction to copy a page instead of mvpg or a couple of mvc instructions. Copying a huge page is 25% faster this way. Also bypass caches when copying pages since only parts of a page will be used afterwards. Especially when copying a huge page this would kick everything out of the L1 and L2 data caches on a zEC12 machine. Signed-off-by: Heiko Carstens

s390/mm: do not initialize storage keys

2013-10-24T15:17:00+00:00

With dirty and referenced bits implemented in software it is unnecessary to initialize the storage key for every page. With this patch not a single storage key operation is done for a system that does not use KVM. For KVM set_pte_at/pgste_set_key will do the initialization for the guest view of the storage key when the mapping for the page is established in the host. Signed-off-by: Martin Schwidefsky

s390/mm: implement software referenced bits

2013-08-29T11:20:11+00:00

The last remaining use for the storage key of the s390 architecture is reference counting. The alternative is to make page table entries invalid while they are old. On access the fault handler marks the pte/pmd as young which makes the pte/pmd valid if the access rights allow read access. The pte/pmd invalidations required for software managed reference bits cost a bit of performance, on the other hand the RRBE/RRBM instructions to read and reset the referenced bits are quite expensive as well. Reviewed-by: Gerald Schaefer Signed-off-by: Martin Schwidefsky

s390/mm: remove dead pfmf inline assembly

2013-08-22T10:20:05+00:00

Signed-off-by: Heiko Carstens Signed-off-by: Martin Schwidefsky