#29 aufs4.19.17+ fails to apply from 4.19.63

[Yoann Ricordel]

From 4.19.63 onwards, with a series of patches using down_read_killable instead of down_read in fs/proc/base.c, the patch from branch aufs4.19.17+ no longer applies:

patching file fs/proc/base.c
Hunk 1 FAILED 2016/2016.
    down_read(&mm->mmap_sem);
    vma = find_exact_vma(mm, vm_start, vm_end);
    if (vma && vma->vm_file) {

-       *path = vma->vm_file->f_path;
+       *path = vma_pr_or_file(vma)->f_path;
        path_get(path);
        rc = 0;
    }

[J. R. Okajima]

"Yoann Ricordel":

From 4.19.63 onwards, with a series of patches using down_read_killable instead of down_read in fs/proc/base.c, the patch from branch aufs4.19.17+ no longer applies:

Thanx for the report.

In linux-v4.19.63, the commit
b07687243d4a 2019-07-31 mm: use down_read_killable for locking
mmap_sem in access_remote_vm
is backported, which is originally added in linux-v5.3-rc1.
1e426fe28261 2019-07-12 mm: use down_read_killable for locking
mmap_sem in access_remote_vm

In other words, aufs5-mmap.patch in aufs5.x-rcN branch has already
addressed the change in mainline.
Then can we apply aufs5.x-rcN#aufs5-mmap.patch onto linux-v4.19.63? I
don't know, but I am sure at least the related part/chunk in the patch
has to be applied.

Anyway I will create and release aufs4.19.63+ branch as soon as
possible.

J. R. Okajima

[J. R. Okajima]

"J. R. Okajima":

Anyway I will create and release aufs4.19.63+ branch as soon as
possible.

I found that the commit is backported into v5.2.5, but not in v4.20 and
v5.1 series.
I will create aufs5.2.5+ too.

J. R. Okajima

[Yoann Ricordel]

Hello,

I could test with Linux 4.19.75 and it's now working perfecfly, thank very much you for the fix :-)

Yoann.