linux-kernel - Re: [PATCH v2 5/6] proc: use down_read_killable mmap_sem for /proc/pid/map

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Thu, 13 Jun 2019 13:32:31 -0700
From:   Andrei Vagin <avagin@...il.com>
To:     Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
Cc:     linux-mm@...ck.org, Andrew Morton <akpm@...ux-foundation.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Oleg Nesterov <oleg@...hat.com>,
        Matthew Wilcox <willy@...radead.org>,
        Michal Hocko <mhocko@...nel.org>,
        Cyrill Gorcunov <gorcunov@...il.com>,
        Kirill Tkhai <ktkhai@...tuozzo.com>,
        Michal Koutný <mkoutny@...e.com>,
        Al Viro <viro@...iv.linux.org.uk>,
        Roman Gushchin <guro@...com>, Dmitry Safonov <dima@...sta.com>,
        Mike Rapoport <rppt@...ux.ibm.com>
Subject: Re: [PATCH v2 5/6] proc: use down_read_killable mmap_sem for /proc/pid/map_files

On Thu, Jun 13, 2019 at 1:15 AM Konstantin Khlebnikov
<khlebnikov@...dex-team.ru> wrote:
>
> On 13.06.2019 2:14, Andrei Vagin wrote:
> > On Sun, Jun 09, 2019 at 01:09:00PM +0300, Konstantin Khlebnikov wrote:
> >> Do not stuck forever if something wrong.
> >> Killable lock allows to cleanup stuck tasks and simplifies investigation.
> >
> > This patch breaks the CRIU project, because stat() returns EINTR instead
> > of ENOENT:
> >
> > [root@...4 criu]# stat /proc/self/map_files/0-0
> > stat: cannot stat '/proc/self/map_files/0-0': Interrupted system call
>
> Good catch.
>
> It seems CRIU tests has good coverage for darkest corners of kernel API.
> Kernel CI projects should use it. I suppose you know how to promote this. =)

I remember Mike was trying to add the CRIU test suite into kernel-ci,
but it looks like this ended up with nothing.

The good thing here is that we have our own kernel-ci:
https://travis-ci.org/avagin/linux/builds

Travis-CI doesn't allow to replace the kernel, so we use CRIU to
dump/restore a ssh session and travis doesn't notice when we kexec a
new kernel.

>
> >
> > Here is one inline comment with the fix for this issue.
> >
> >>
> >> It seems ->d_revalidate() could return any error (except ECHILD) to
> >> abort validation and pass error as result of lookup sequence.
> >>
> >> Signed-off-by: Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
> >> Reviewed-by: Roman Gushchin <guro@...com>
> >> Reviewed-by: Cyrill Gorcunov <gorcunov@...il.com>
> >> Reviewed-by: Kirill Tkhai <ktkhai@...tuozzo.com>
> >
> > It was nice to see all four of you in one place :).
> >
> >> Acked-by: Michal Hocko <mhocko@...e.com>
> >> ---
> >>   fs/proc/base.c |   27 +++++++++++++++++++++------
> >>   1 file changed, 21 insertions(+), 6 deletions(-)
> >>
> >> diff --git a/fs/proc/base.c b/fs/proc/base.c
> >> index 9c8ca6cd3ce4..515ab29c2adf 100644
> >> --- a/fs/proc/base.c
> >> +++ b/fs/proc/base.c
> >> @@ -1962,9 +1962,12 @@ static int map_files_d_revalidate(struct dentry *dentry, unsigned int flags)
> >>              goto out;
> >>
> >>      if (!dname_to_vma_addr(dentry, &vm_start, &vm_end)) {
> >> -            down_read(&mm->mmap_sem);
> >> -            exact_vma_exists = !!find_exact_vma(mm, vm_start, vm_end);
> >> -            up_read(&mm->mmap_sem);
> >> +            status = down_read_killable(&mm->mmap_sem);
> >> +            if (!status) {
> >> +                    exact_vma_exists = !!find_exact_vma(mm, vm_start,
> >> +                                                        vm_end);
> >> +                    up_read(&mm->mmap_sem);
> >> +            }
> >>      }
> >>
> >>      mmput(mm);
> >> @@ -2010,8 +2013,11 @@ static int map_files_get_link(struct dentry *dentry, struct path *path)
> >>      if (rc)
> >>              goto out_mmput;
> >>
> >> +    rc = down_read_killable(&mm->mmap_sem);
> >> +    if (rc)
> >> +            goto out_mmput;
> >> +
> >>      rc = -ENOENT;
> >> -    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;
> >> @@ -2107,7 +2113,10 @@ static struct dentry *proc_map_files_lookup(struct inode *dir,
> >>      if (!mm)
> >>              goto out_put_task;
> >>
> >> -    down_read(&mm->mmap_sem);
> >> +    result = ERR_PTR(-EINTR);
> >> +    if (down_read_killable(&mm->mmap_sem))
> >> +            goto out_put_mm;
> >> +
> >
> >       result = ERR_PTR(-ENOENT);
> >
> >>      vma = find_exact_vma(mm, vm_start, vm_end);
> >>      if (!vma)
> >>              goto out_no_vma;
> >> @@ -2118,6 +2127,7 @@ static struct dentry *proc_map_files_lookup(struct inode *dir,
> >>
> >>   out_no_vma:
> >>      up_read(&mm->mmap_sem);
> >> +out_put_mm:
> >>      mmput(mm);
> >>   out_put_task:
> >>      put_task_struct(task);
> >> @@ -2160,7 +2170,12 @@ proc_map_files_readdir(struct file *file, struct dir_context *ctx)
> >>      mm = get_task_mm(task);
> >>      if (!mm)
> >>              goto out_put_task;
> >> -    down_read(&mm->mmap_sem);
> >> +
> >> +    ret = down_read_killable(&mm->mmap_sem);
> >> +    if (ret) {
> >> +            mmput(mm);
> >> +            goto out_put_task;
> >> +    }
> >>
> >>      nr_files = 0;
> >>