lkml.org 
[lkml]   [2012]   [Jun]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH -V8 05/16] hugetlb: avoid taking i_mmap_mutex in unmap_single_vma() for hugetlb
    On Sat, Jun 09, 2012 at 06:33:05PM +0530, Aneesh Kumar K.V wrote:
    > Johannes Weiner <hannes@cmpxchg.org> writes:
    >
    > > On Sat, Jun 09, 2012 at 02:29:50PM +0530, Aneesh Kumar K.V wrote:
    > >> From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
    > >>
    > >> i_mmap_mutex lock was added in unmap_single_vma by 502717f4e ("hugetlb:
    > >> fix linked list corruption in unmap_hugepage_range()") but we don't use
    > >> page->lru in unmap_hugepage_range any more. Also the lock was taken
    > >> higher up in the stack in some code path. That would result in deadlock.
    > >>
    > >> unmap_mapping_range (i_mmap_mutex)
    > >> -> unmap_mapping_range_tree
    > >> -> unmap_mapping_range_vma
    > >> -> zap_page_range_single
    > >> -> unmap_single_vma
    > >> -> unmap_hugepage_range (i_mmap_mutex)
    > >>
    > >> For shared pagetable support for huge pages, since pagetable pages are ref
    > >> counted we don't need any lock during huge_pmd_unshare. We do take
    > >> i_mmap_mutex in huge_pmd_share while walking the vma_prio_tree in mapping.
    > >> (39dde65c9940c97f ("shared page table for hugetlb page")).
    > >>
    > >> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
    > >
    > > This patch (together with the previous one) seems like a bugfix that's
    > > not really related to the hugetlb controller, unless I miss something.
    > >
    > > Could you please submit the fix separately?
    >
    > Patches upto 6 can really got in a separate series. I was not sure
    > whether I should split them. I will post that as a separate series now

    Ok, thanks, that will make it easier to upstream the controller.

    > > Maybe also fold the two patches into one and make it a single bugfix
    > > change that gets rid of the lock by switching away from page->lru.
    >
    > I wanted to make sure the patch that drop i_mmap_mutex is a separate one
    > so that we understand and document the locking details separately

    Nothing prevents you from writing a proper changelog :-) But changing
    from page->lru to an on-stack array does not have any merit by itself,
    so it just seems like a needless dependency between two patches that
    fix one problem (pita for backports into stable/distro kernels).


    \
     
     \ /
      Last update: 2012-06-09 17:21    [W:0.025 / U:0.308 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site