On Wed, 15 Dec 2010, Sage Weil wrote: > From: Henry C Chang <[email protected]> > > For read operation, we have to set the argument _write_ of get_user_pages > to 1 since we will write data to pages. Also, we need to SetPageDirty before > releasing these pages.
I made one small change from your version (see below). That aside, I have no idea if this patch is the right thing to do. I thought the page bits were mainly important for the page cache... what do we accomplish by marking user pages dirty this way? sage > > Signed-off-by: Henry C Chang <[email protected]> > Signed-off-by: Sage Weil <[email protected]> > --- > fs/ceph/file.c | 4 ++-- > include/linux/ceph/libceph.h | 3 ++- > net/ceph/pagevec.c | 6 +++--- > 3 files changed, 7 insertions(+), 6 deletions(-) > > diff --git a/fs/ceph/file.c b/fs/ceph/file.c > index e860d8f..e0f136d 100644 > --- a/fs/ceph/file.c > +++ b/fs/ceph/file.c > @@ -384,7 +384,7 @@ static ssize_t ceph_sync_read(struct file *file, char > __user *data, > > if (file->f_flags & O_DIRECT) { > num_pages = calc_pages_for((unsigned long)data, len); > - pages = ceph_get_direct_page_vector(data, num_pages); > + pages = ceph_get_direct_page_vector(data, num_pages, true); > } else { > num_pages = calc_pages_for(off, len); > pages = ceph_alloc_page_vector(num_pages, GFP_NOFS); > @@ -522,7 +522,7 @@ more: > return -ENOMEM; > > if (file->f_flags & O_DIRECT) { > - pages = ceph_get_direct_page_vector(data, num_pages); > + pages = ceph_get_direct_page_vector(data, num_pages, false); > if (IS_ERR(pages)) { > ret = PTR_ERR(pages); > goto out; > diff --git a/include/linux/ceph/libceph.h b/include/linux/ceph/libceph.h > index 9e76d35..11992e8 100644 > --- a/include/linux/ceph/libceph.h > +++ b/include/linux/ceph/libceph.h > @@ -227,7 +227,8 @@ extern int ceph_open_session(struct ceph_client *client); > extern void ceph_release_page_vector(struct page **pages, int num_pages); > > extern struct page **ceph_get_direct_page_vector(const char __user *data, > - int num_pages); > + int num_pages, > + bool write_page); > extern void ceph_put_page_vector(struct page **pages, int num_pages); > extern void ceph_release_page_vector(struct page **pages, int num_pages); > extern struct page **ceph_alloc_page_vector(int num_pages, gfp_t flags); > diff --git a/net/ceph/pagevec.c b/net/ceph/pagevec.c > index ac34fee..128d5ba 100644 > --- a/net/ceph/pagevec.c > +++ b/net/ceph/pagevec.c > @@ -13,18 +13,18 @@ > * build a vector of user pages > */ > struct page **ceph_get_direct_page_vector(const char __user *data, > - int num_pages) > + int num_pages, bool write_page) > { > struct page **pages; > int rc; > > - pages = kmalloc(sizeof(*pages) * num_pages, GFP_NOFS); > + pages = kzalloc(sizeof(*pages) * num_pages, GFP_NOFS); > if (!pages) > return ERR_PTR(-ENOMEM); > > down_read(¤t->mm->mmap_sem); > rc = get_user_pages(current, current->mm, (unsigned long)data, > - num_pages, 0, 0, pages, NULL); > + num_pages, write_page, 0, pages, NULL); > up_read(¤t->mm->mmap_sem); > if (rc < 0) Your version had this as if (rc < num_pages) but as far as I can see get_user_pages always returns either the full result, or nothing (the __get_user_pages fail path releases what it has so far). Hmm, that actually means the kzalloc change isn't needed too, right? > goto fail; > -- > 1.7.1 > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to [email protected] > More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
