On 07/27/2015 12:13 AM, Michal Hocko wrote:
On Fri 24-07-15 10:27:02, Mike Kravetz wrote:
On 07/23/2015 03:18 PM, kbuild test robot wrote:
tree: git://git.cmpxchg.org/linux-mmotm.git master
head: 61f5f835b6f06fbc233481b5d3c0afd71ecf54e8
commit: 0c5e1e8ed55974975bb829e4b93cf19aa0dfcafc [229/385] hugetlbfs: add hugetlbfs_fallocate()
config: i386-randconfig-r0-201529 (attached as .config)
reproduce:
git checkout 0c5e1e8ed55974975bb829e4b93cf19aa0dfcafc
# save the attached .config to linux build tree
make ARCH=i386
All error/warnings (new ones prefixed by >>):
fs/hugetlbfs/inode.c: In function 'hugetlbfs_fallocate':
fs/hugetlbfs/inode.c:578:13: error: 'struct vm_area_struct' has no member named 'vm_policy'
pseudo_vma.vm_policy =
^
fs/hugetlbfs/inode.c:579:4: error: implicit declaration of function 'mpol_shared_policy_lookup' [-Werror=implicit-function-declaration]
mpol_shared_policy_lookup(&HUGETLBFS_I(inode)->policy,
^
fs/hugetlbfs/inode.c:595:28: error: 'struct vm_area_struct' has no member named 'vm_policy'
mpol_cond_put(pseudo_vma.vm_policy);
^
fs/hugetlbfs/inode.c:601:27: error: 'struct vm_area_struct' has no member named 'vm_policy'
mpol_cond_put(pseudo_vma.vm_policy);
^
cc1: some warnings being treated as errors
vim +578 fs/hugetlbfs/inode.c
572 if (signal_pending(current)) {
573 error = -EINTR;
574 break;
575 }
576
577 /* Get policy based on index */
> 578 pseudo_vma.vm_policy =
> 579 mpol_shared_policy_lookup(&HUGETLBFS_I(inode)->policy,
580 index);
581
582 /* addr is the offset within the file (zero based) */
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
Michal already added a patch to mmotm. The patch below is functionally
equivalent but moves the #ifdef out of the executable code path, and
modifies a comment. This has been functional/stress tested in a kernel
without CONFIG_NUMA defined.
I will drop my quick hack once Andrew picks your patch which is indeed
better.
One nit below...
hugetlbfs: build fix fallocate if not CONFIG_NUMA
When fallocate preallocation allocates pages, it will use the
defined numa policy. However, if numa is not defined there is
no such policy and no code should reference numa policy. Create
wrappers to isolate policy manipulation code that are NOOP in
the non-NUMA case.
Signed-off-by: Mike Kravetz <mike.kravetz@xxxxxxxxxx>
---
fs/hugetlbfs/inode.c | 39 ++++++++++++++++++++++++++++++---------
1 file changed, 30 insertions(+), 9 deletions(-)
diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index d977cae..4bae359 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -85,6 +85,29 @@ static const match_table_t tokens = {
{Opt_err, NULL},
};
+#ifdef CONFIG_NUMA
+static inline void hugetlb_set_vma_policy(struct vm_area_struct *vma,
+ struct inode *inode, pgoff_t index)
+{
+ vma->vm_policy = mpol_shared_policy_lookup(&HUGETLBFS_I(inode)->policy,
+ index);
+}
+
+static inline void hugetlb_vma_mpol_cond_put(struct vm_area_struct *vma)
The naming could be better. What about hugetlb_drop_vma_policy to be
symmetric to hugetlb_set_vma_policy. Or is there any reason to expose
that this is a cond_put?
I like the name. Will send an updated patch with the change.
--
Mike Kravetz
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>