Re: [PATCH v5 3/5] vmcore: Introduce remap_oldmem_pfn_range()

[HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>]

(2013/06/12 18:13), Michael Holzheu wrote:
> On Tue, 11 Jun 2013 21:42:15 +0900
> HATAYAMA Daisuke <d.hatayama@gmail.com> wrote:
>
>> 2013/6/11 Michael Holzheu <holzheu@linux.vnet.ibm.com>:
>>> On Mon, 10 Jun 2013 22:40:24 +0900
>>> HATAYAMA Daisuke <d.hatayama@gmail.com> wrote:
>>>
>>>> 2013/6/8 Michael Holzheu <holzheu@linux.vnet.ibm.com>:
<cut>
> static int mmap_vmcore_fault(struct vm_area_struct *vma, struct'vm_fault *vmf)
> {
>          struct address_space *mapping = vma->vm_private_data;
>          pgoff_t index = vmf->pgoff;
>          struct page *page;
>          loff_t src;
>          char *buf;
>
>          page = find_or_create_page(mapping, index, GFP_KERNEL);
>          if (!page)
>                  return VM_FAULT_OOM;
>          if (!PageUptodate(page)) {
>                  src = index << PAGE_CACHE_SHIFT;
>                  buf = (void *) (page_to_pfn(page) << PAGE_SHIFT);
>                  if (__read_vmcore(buf, PAGE_SIZE, &src, 0) < 0) {
>                          unlock_page(page);
>                          return VM_FAULT_SIGBUS;
>                  }
>                  SetPageUptodate(page);
>          }
>          unlock_page(page);
>          vmf->page = page;
>          return 0;
> }
>
> Perhaps one open issue remains:
>
> Can we remove the page from the page cache if __read_vmcore() fails?
>

Yes, use page_cache_release() after unlocking the page like:

                    if (__read_vmcore(buf, PAGE_SIZE, &src, 0) < 0) {
                            unlock_page(page);
+                          page_cache_release(page);
                            return VM_FAULT_SIGBUS;
                    }

BTW, you now keep file->f_mapping in vma->vm_private_data, but the vma already has the file object in its vma->vm_file member. You can get the mapping by vma->vm_file->f_mapping without necessity of vma->vm_private_data.

-- 
Thanks.
HATAYAMA, Daisuke