Re: [PATCH] umh: always return error when helper was not called

[Sergei Trofimovich <slyfox@gentoo.org>]

On Thu, 16 Apr 2020 05:11:27 +0000
Luis Chamberlain <mcgrof@kernel.org> wrote:

> Sergei, first, thanks for your patch and bug report!!
> 
> On Wed, Apr 15, 2020 at 07:59:40AM +0100, Sergei Trofimovich wrote:
> > Before this change on a system with the following setup crashed kernel:
> > 
> > ```
> > CONFIG_STATIC_USERMODEHELPER=y
> > CONFIG_STATIC_USERMODEHELPER_PATH=""
> > kernel.core_pattern = |/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %h %e
> > ```  
> 
> Let us backtrack. The combination of:
> 
> CONFIG_STATIC_USERMODEHELPER=y                                                
> CONFIG_STATIC_USERMODEHELPER_PATH=""
> 
> is documented on the kconfig files for when you *want to disable all
> usermode helper programs.
> 
> > The crash happens when a core dump is attempted:
> > 
> > ```
> > [    2.819676] BUG: kernel NULL pointer dereference, address: 0000000000000020
> > [    2.819859] #PF: supervisor read access in kernel mode
> > [    2.820035] #PF: error_code(0x0000) - not-present page
> > [    2.820188] PGD 0 P4D 0
> > [    2.820305] Oops: 0000 [#1] SMP PTI
> > [    2.820436] CPU: 2 PID: 89 Comm: a Not tainted 5.7.0-rc1+ #7
> > [    2.820680] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ?-20190711_202441-buildvm-armv7-10.arm.fedoraproject.org-2.fc31 04/01/2014
> > [    2.821150] RIP: 0010:do_coredump+0xd80/0x1060
> > [    2.821385] Code: e8 95 11 ed ff 48 c7 c6 cc a7 b4 81 48 8d bd 28 ff ff ff 89 c2 e8 70 f1 ff ff 41 89 c2 85 c0 0f 84 72 f7 ff ff e9 b4 fe ff ff <48> 8b 57 20 0f b7 02 66 25 00 f0 66 3d 00 8
> > 0 0f 84 9c 01 00 00 44
> > [    2.822014] RSP: 0000:ffffc9000029bcb8 EFLAGS: 00010246
> > [    2.822339] RAX: 0000000000000000 RBX: ffff88803f860000 RCX: 000000000000000a
> > [    2.822746] RDX: 0000000000000009 RSI: 0000000000000282 RDI: 0000000000000000
> > [    2.823141] RBP: ffffc9000029bde8 R08: 0000000000000000 R09: ffffc9000029bc00
> > [    2.823508] R10: 0000000000000001 R11: ffff88803dec90be R12: ffffffff81c39da0
> > [    2.823902] R13: ffff88803de84400 R14: 0000000000000000 R15: 0000000000000000
> > [    2.824285] FS:  00007fee08183540(0000) GS:ffff88803e480000(0000) knlGS:0000000000000000
> > [    2.824767] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > [    2.825111] CR2: 0000000000000020 CR3: 000000003f856005 CR4: 0000000000060ea0
> > [    2.825479] Call Trace:
> > [    2.825790]  get_signal+0x11e/0x720
> > [    2.826087]  do_signal+0x1d/0x670
> > [    2.826361]  ? force_sig_info_to_task+0xc1/0xf0
> > [    2.826691]  ? force_sig_fault+0x3c/0x40
> > [    2.826996]  ? do_trap+0xc9/0x100
> > [    2.827179]  exit_to_usermode_loop+0x49/0x90
> > [    2.827359]  prepare_exit_to_usermode+0x77/0xb0
> > [    2.827559]  ? invalid_op+0xa/0x30
> > [    2.827747]  ret_from_intr+0x20/0x20
> > [    2.827921] RIP: 0033:0x55e2c76d2129
> > [    2.828107] Code: 2d ff ff ff e8 68 ff ff ff 5d c6 05 18 2f 00 00 01 c3 0f 1f 80 00 00 00 00 c3 0f 1f 80 00 00 00 00 e9 7b ff ff ff 55 48 89 e5 <0f> 0b b8 00 00 00 00 5d c3 66 2e 0f 1f 84 0
> > 0 00 00 00 00 0f 1f 40
> > [    2.828603] RSP: 002b:00007fffeba5e080 EFLAGS: 00010246
> > [    2.828801] RAX: 000055e2c76d2125 RBX: 0000000000000000 RCX: 00007fee0817c718
> > [    2.829034] RDX: 00007fffeba5e188 RSI: 00007fffeba5e178 RDI: 0000000000000001
> > [    2.829257] RBP: 00007fffeba5e080 R08: 0000000000000000 R09: 00007fee08193c00
> > [    2.829482] R10: 0000000000000009 R11: 0000000000000000 R12: 000055e2c76d2040
> > [    2.829727] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
> > [    2.829964] CR2: 0000000000000020
> > [    2.830149] ---[ end trace ceed83d8c68a1bf1 ]---
> > ```
> > 
> > Here is the sequence of events why it happens:
> > fs/coredump.c:do_coredump():
> > 1. create 'coredump_params = { .file = NULL }'
> > 2. detect pipe mode
> > 3. `call_usermodehelper_setup(..., umh_pipe_setup, ...)`
> > 4. `call_usermodehelper_exec()`
> > 5. (if both succeeded) `file_start_write(cprm.file);`
> > 
> > Here crash happens at [5.] as `cprm.file` is still NULL.
> > 
> > Normally it works because `fs/coredump.c:umh_pipe_setup()` is called
> > successfully and populates `.file` field (or returns the error):
> > 
> > ```
> > static int umh_pipe_setup(struct subprocess_info *info, struct cred *new)
> > {
> >         //...
> >         struct coredump_params *cp = (struct coredump_params *)info->data;
> >         // ...
> >         cp->file = files[1];
> >         // ...
> > }
> > ```
> > 
> > But in our case neither happens because `kernel/umh.c:call_usermodehelper_exec()`
> > has a special case:
> > 
> > ```
> > int call_usermodehelper_exec(struct subprocess_info *sub_info, int wait)
> > {
> >     int retval = 0;
> >     // ...
> >     /*
> >      * If there is no binary for us to call, then just return and get out of
> >      * here.  This allows us to set STATIC_USERMODEHELPER_PATH to "" and
> >      * disable all call_usermodehelper() calls.
> >      */
> >     if (strlen(sub_info->path) == 0)
> >         goto out;
> >     ...
> >     out:
> >         // ...
> >         return retval;
> > 
> > ```
> > 
> > This breaks assumption of `do_coredump()`: "either helper was called successfully
> > and created a file to dump core to or it failed".
> > 
> > This change converts this special case to `-EPERM` error.
> > 
> > This way we notify user that helper call was not successful
> > and don't attempt to act on uninitialized `.file` field.
> > 
> > User gets `"Core dump to |%s pipe failed\n` dmesg entry.
> > 
> > Reported-by: Sergey Kvachonok <ravenexp@gmail.com>
> > Reported-by: Tony Vroon <chainsaw@gentoo.org>
> > Bug: https://bugzilla.kernel.org/show_bug.cgi?id=199795
> > Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
> > CC: Luis Chamberlain <mcgrof@kernel.org>
> > ---
> >  kernel/umh.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> > 
> > diff --git a/kernel/umh.c b/kernel/umh.c
> > index 7f255b5a8845..66b02634a9ba 100644
> > --- a/kernel/umh.c
> > +++ b/kernel/umh.c
> > @@ -565,8 +565,10 @@ int call_usermodehelper_exec(struct subprocess_info *sub_info, int wait)
> >  	 * here.  This allows us to set STATIC_USERMODEHELPER_PATH to "" and
> >  	 * disable all call_usermodehelper() calls.
> >  	 */
> > -	if (strlen(sub_info->path) == 0)
> > +	if (strlen(sub_info->path) == 0) {
> > +		retval = -EPERM;
> >  		goto out;
> > +	}
> >  
> >  	/*
> >  	 * Set the completion pointer only if there is a waiter.  
> 
> Stakeholders with this setup likely already are relying on the fact that
> we don't return -EPERM. Your change to return -EPERM may fix the crash
> you are seeing, but it may also break existing userspace, granted
> for a crashdump that may not matter much, however I agree it is stupid
> to crash on a crash :)
> 
> Anyway, I don't think its a good idea to return -EPERM unless
> stakeholders really find a strong reason to change old behaviour.

I agree changing existing API semantics is dangerous.
Maybe tweaking a docstring to call_usermodehelper_exec() would be helpful.
Something like:

--- a/kernel/umh.c
+++ b/kernel/umh.c
@@ -544,6 +544,11 @@ EXPORT_SYMBOL_GPL(fork_usermode_blob);
  * Runs a user-space application.  The application is started
  * asynchronously if wait is not set, and runs as a child of system workqueues.
  * (ie. it runs with full root capabilities and optimized affinity).
+ *
+ * Note: successful return value does not guarantee helper was called at all.
+ * You can't rely on sub_info->{init,cleanup} being called even for UMH_WAIT_*
+ * wait modes as STATIC_USERMODEHELPER_PATH="" turns all helpers
+ * into a successful no-op.
  */
 int call_usermodehelper_exec(struct subprocess_info *sub_info, int wait)
 {

> Can you try this patch instead?

Gave it a light crash-test locally. Works fine!

Tested-by: Sergei Trofimovich <slyfox@gentoo.org>

> diff --git a/fs/coredump.c b/fs/coredump.c
> index f8296a82d01d..6957d513685e 100644
> --- a/fs/coredump.c
> +++ b/fs/coredump.c
> @@ -786,6 +786,15 @@ void do_coredump(const kernel_siginfo_t *siginfo)
>  	if (displaced)
>  		put_files_struct(displaced);
>  	if (!dump_interrupted()) {
> +		/*
> +		 * umh disabled with CONFIG_STATIC_USERMODEHELPER_PATH="" would
> +		 * have this set to NULL.
> +		 */
> +		if (!cprm.file) {
> +			printk(KERN_INFO "Core dump to |%s disabled\n",
> +			       cn.corename);
> +			goto close_fail;
> +		}
>  		file_start_write(cprm.file);
>  		core_dumped = binfmt->core_dump(&cprm);
>  		file_end_write(cprm.file);


-- 

  Sergei