[PATCH] kmod: make request_module() return an error when autoloading is disabled

[PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

It's long been possible to disable kernel module autoloading completely
by setting /proc/sys/kernel/modprobe to the empty string.  This can be
preferable to setting it to a nonexistent file since it avoids the
overhead of an attempted execve(), avoids potential deadlocks, and
avoids the call to security_kernel_module_request() and thus on
SELinux-based systems eliminates the need to write SELinux rules to
dontaudit module_request.

However, when module autoloading is disabled in this way,
request_module() returns 0.  This is broken because callers expect 0 to
mean that the module was successfully loaded.

Apparently this was never noticed because this method of disabling
module autoloading isn't used much, and also most callers don't use the
return value of request_module() since it's always necessary to check
whether the module registered its functionality or not anyway.  But
improperly returning 0 can indeed confuse a few callers, for example
get_fs_type() in fs/filesystems.c where it causes a WARNING to be hit:

	if (!fs && (request_module("fs-%.*s", len, name) == 0)) {
		fs = __get_fs_type(name, len);
		WARN_ONCE(!fs, "request_module fs-%.*s succeeded, but still no fs?\n", len, name);
	}

This is easily reproduced with:

	echo > /proc/sys/kernel/modprobe
	mount -t NONEXISTENT none /

It causes:

	request_module fs-NONEXISTENT succeeded, but still no fs?
	WARNING: CPU: 1 PID: 1106 at fs/filesystems.c:275 get_fs_type+0xd6/0xf0
	[...]

Arguably this warning is broken and should be removed, since the module
could have been unloaded already.  However, request_module() should also
correctly return an error when it fails.  So let's make it return
-ENOENT, which matches the error when the modprobe binary doesn't exist.

Cc: stable@vger.kernel.org
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jeff Vander Stoep <jeffv@google.com>
Cc: Jessica Yu <jeyu@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
---
 kernel/kmod.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/kmod.c b/kernel/kmod.c
index bc6addd9152b..a2de58de6ab6 100644
--- a/kernel/kmod.c
+++ b/kernel/kmod.c
@@ -120,7 +120,7 @@ static int call_modprobe(char *module_name, int wait)
  * invoke it.
  *
  * If module auto-loading support is disabled then this function
- * becomes a no-operation.
+ * simply returns -ENOENT.
  */
 int __request_module(bool wait, const char *fmt, ...)
 {
@@ -137,7 +137,7 @@ int __request_module(bool wait, const char *fmt, ...)
 	WARN_ON_ONCE(wait && current_is_async());
 
 	if (!modprobe_path[0])
-		return 0;
+		return -ENOENT;
 
 	va_start(args, fmt);
 	ret = vsnprintf(module_name, MODULE_NAME_LEN, fmt, args);
-- 
2.25.1.481.gfbce0eb801-goog

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> From: Eric Biggers <ebiggers@google.com>
> 
> It's long been possible to disable kernel module autoloading completely
> by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> preferable

preferable but ... not documented. Or was this documented or recommended
somewhere?

> to setting it to a nonexistent file since it avoids the
> overhead of an attempted execve(), avoids potential deadlocks, and
> avoids the call to security_kernel_module_request() and thus on
> SELinux-based systems eliminates the need to write SELinux rules to
> dontaudit module_request.
> 
> However, when module autoloading is disabled in this way,
> request_module() returns 0.  This is broken because callers expect 0 to
> mean that the module was successfully loaded.

However this is implicitly not true. For instance, as Neil recently
chased down -- blacklisting a module today returns 0 as well, and so
this corner case is implicitly set to return 0.

> Apparently this was never noticed because this method of disabling
> module autoloading isn't used much, and also most callers don't use the
> return value of request_module() since it's always necessary to check
> whether the module registered its functionality or not anyway.

Right, the de-facto practice of verification of a module to be loaded is
for each caller to ensure with whatever heuristic it needs to ensure the
module is loaded.

> But
> improperly returning 0 can indeed confuse a few callers, for example
> get_fs_type() in fs/filesystems.c where it causes a WARNING to be hit:
> 
> 	if (!fs && (request_module("fs-%.*s", len, name) == 0)) {
> 		fs = __get_fs_type(name, len);
> 		WARN_ONCE(!fs, "request_module fs-%.*s succeeded, but still no fs?\n", len, name);
> 	}
> 
> This is easily reproduced with:
> 
> 	echo > /proc/sys/kernel/modprobe
> 	mount -t NONEXISTENT none /
> 
> It causes:
> 
> 	request_module fs-NONEXISTENT succeeded, but still no fs?
> 	WARNING: CPU: 1 PID: 1106 at fs/filesystems.c:275 get_fs_type+0xd6/0xf0
> 	[...]

Thanks for reporting this.

> Arguably this warning is broken and should be removed, since the module
> could have been unloaded already.

No, the warning is present *because* debuggins issues for when the
module which did not load is a rootfs is *really* hard to debug. Then,
if the culprit of the issue is a userspace modprobe bug (it happens)
this makes debugging *very* difficult as you won't know what failed at
all, you just get a silent failed boot.

> However, request_module() should also
> correctly return an error when it fails.  So let's make it return
> -ENOENT, which matches the error when the modprobe binary doesn't exist.

This is a user experience change though, and I wouldn't have on my radar
who would use this, and expects the old behaviour. Josh, would you by
chance?

I'd like this to be more an RFC first so we get vetted parties to
review. I take it this and Neil's case are cases we should revisit now,
properly document as we didn't before, ensure we don't break anything,
and also extend the respective kmod selftests to ensure we don't break
these corner cases in the future.

> Cc: stable@vger.kernel.org
> Cc: Alexei Starovoitov <ast@kernel.org>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Cc: Jeff Vander Stoep <jeffv@google.com>
> Cc: Jessica Yu <jeyu@kernel.org>
> Cc: Kees Cook <keescook@chromium.org>
> Cc: Luis Chamberlain <mcgrof@kernel.org>
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> ---
>  kernel/kmod.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/kmod.c b/kernel/kmod.c
> index bc6addd9152b..a2de58de6ab6 100644
> --- a/kernel/kmod.c
> +++ b/kernel/kmod.c
> @@ -120,7 +120,7 @@ static int call_modprobe(char *module_name, int wait)
>   * invoke it.
>   *
>   * If module auto-loading support is disabled then this function
> - * becomes a no-operation.
> + * simply returns -ENOENT.
>   */
>  int __request_module(bool wait, const char *fmt, ...)
>  {
> @@ -137,7 +137,7 @@ int __request_module(bool wait, const char *fmt, ...)
>  	WARN_ON_ONCE(wait && current_is_async());
>  
>  	if (!modprobe_path[0])
> -		return 0;
> +		return -ENOENT;
>  
>  	va_start(args, fmt);
>  	ret = vsnprintf(module_name, MODULE_NAME_LEN, fmt, args);
> -- 
> 2.25.1.481.gfbce0eb801-goog
> 

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > From: Eric Biggers <ebiggers@google.com>
> > 
> > It's long been possible to disable kernel module autoloading completely
> > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > preferable
> 
> preferable but ... not documented. Or was this documented or recommended
> somewhere?
> 
> > to setting it to a nonexistent file since it avoids the
> > overhead of an attempted execve(), avoids potential deadlocks, and
> > avoids the call to security_kernel_module_request() and thus on
> > SELinux-based systems eliminates the need to write SELinux rules to
> > dontaudit module_request.

Not that I know of, though I didn't look too hard.  proc(5) mentions
/proc/sys/kernel/modprobe but doesn't mention the empty string case.

In any case, it's been supported for a long time, and it's useful for the
reasons I mentioned.

> > 
> > However, when module autoloading is disabled in this way,
> > request_module() returns 0.  This is broken because callers expect 0 to
> > mean that the module was successfully loaded.
> 
> However this is implicitly not true. For instance, as Neil recently
> chased down -- blacklisting a module today returns 0 as well, and so
> this corner case is implicitly set to return 0.

That sounds like another similar bug, but in the modprobe program instead of in
the kernel.  Do you have a link to the discussion about it?

> 
> > But
> > improperly returning 0 can indeed confuse a few callers, for example
> > get_fs_type() in fs/filesystems.c where it causes a WARNING to be hit:
> > 
> > 	if (!fs && (request_module("fs-%.*s", len, name) == 0)) {
> > 		fs = __get_fs_type(name, len);
> > 		WARN_ONCE(!fs, "request_module fs-%.*s succeeded, but still no fs?\n", len, name);
> > 	}
> > 
> > This is easily reproduced with:
> > 
> > 	echo > /proc/sys/kernel/modprobe
> > 	mount -t NONEXISTENT none /
> > 
> > It causes:
> > 
> > 	request_module fs-NONEXISTENT succeeded, but still no fs?
> > 	WARNING: CPU: 1 PID: 1106 at fs/filesystems.c:275 get_fs_type+0xd6/0xf0
> > 	[...]
> 
> Thanks for reporting this.
> 
> > Arguably this warning is broken and should be removed, since the module
> > could have been unloaded already.
> 
> No, the warning is present *because* debuggins issues for when the
> module which did not load is a rootfs is *really* hard to debug. Then,
> if the culprit of the issue is a userspace modprobe bug (it happens)
> this makes debugging *very* difficult as you won't know what failed at
> all, you just get a silent failed boot.

I meant that it's broken to use WARN_ON(), because it's a userspace triggerable
condition.  WARN_ON() is for kernel bugs only.  Of course, if it's a useful
warning, it can still be left in as pr_warn().

> > However, request_module() should also
> > correctly return an error when it fails.  So let's make it return
> > -ENOENT, which matches the error when the modprobe binary doesn't exist.
> 
> This is a user experience change though, and I wouldn't have on my radar
> who would use this, and expects the old behaviour. Josh, would you by
> chance?
> 
> I'd like this to be more an RFC first so we get vetted parties to
> review. I take it this and Neil's case are cases we should revisit now,
> properly document as we didn't before, ensure we don't break anything,
> and also extend the respective kmod selftests to ensure we don't break
> these corner cases in the future.

This patch only affects kernel internals, not the userspace API.  So I don't see
why it would be controversial?  I already went through all callers of
request_module() that check its return value, and they all appear to work better
with -ENOENT, since they assume that 0 means the module was loaded.

Incorrectly returning 0 typically causes unnecessary work (checking again
whether the module's functionality is available) or misleading log messages.  In
fact, I can't think of a situation where kernel code would *want* 0 returned in
this case, as it's ambiguous with the module being successfully loaded.

Sure, I'll check whether it would be possible to add a test for this case in
lib/test_kmod.c and tools/testing/selftests/kmod/.

- Eric

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Tue, Mar 10, 2020 at 10:26:20PM -0700, Eric Biggers wrote:
> On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > From: Eric Biggers <ebiggers@google.com>
> > > 
> > > It's long been possible to disable kernel module autoloading completely
> > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > preferable
> > 
> > preferable but ... not documented. Or was this documented or recommended
> > somewhere?
> > 
> > > to setting it to a nonexistent file since it avoids the
> > > overhead of an attempted execve(), avoids potential deadlocks, and
> > > avoids the call to security_kernel_module_request() and thus on
> > > SELinux-based systems eliminates the need to write SELinux rules to
> > > dontaudit module_request.
> 
> Not that I know of, though I didn't look too hard.  proc(5) mentions
> /proc/sys/kernel/modprobe but doesn't mention the empty string case.
> 
> In any case, it's been supported for a long time, and it's useful for the
> reasons I mentioned.

Sure. I think then its important to document it as such then, or perhaps
make a kconfig option which sets this to empty and document it on the
kconfig entry.

> > > However, when module autoloading is disabled in this way,
> > > request_module() returns 0.  This is broken because callers expect 0 to
> > > mean that the module was successfully loaded.
> > 
> > However this is implicitly not true. For instance, as Neil recently
> > chased down -- blacklisting a module today returns 0 as well, and so
> > this corner case is implicitly set to return 0.
> 
> That sounds like another similar bug, but in the modprobe program instead of in
> the kernel.  Do you have a link to the discussion about it?

Nothing public yet AFAICT.

> > > But
> > > improperly returning 0 can indeed confuse a few callers, for example
> > > get_fs_type() in fs/filesystems.c where it causes a WARNING to be hit:
> > > 
> > > 	if (!fs && (request_module("fs-%.*s", len, name) == 0)) {
> > > 		fs = __get_fs_type(name, len);
> > > 		WARN_ONCE(!fs, "request_module fs-%.*s succeeded, but still no fs?\n", len, name);
> > > 	}
> > > 
> > > This is easily reproduced with:
> > > 
> > > 	echo > /proc/sys/kernel/modprobe
> > > 	mount -t NONEXISTENT none /
> > > 
> > > It causes:
> > > 
> > > 	request_module fs-NONEXISTENT succeeded, but still no fs?
> > > 	WARNING: CPU: 1 PID: 1106 at fs/filesystems.c:275 get_fs_type+0xd6/0xf0
> > > 	[...]
> > 
> > Thanks for reporting this.
> > 
> > > Arguably this warning is broken and should be removed, since the module
> > > could have been unloaded already.
> > 
> > No, the warning is present *because* debuggins issues for when the
> > module which did not load is a rootfs is *really* hard to debug. Then,
> > if the culprit of the issue is a userspace modprobe bug (it happens)
> > this makes debugging *very* difficult as you won't know what failed at
> > all, you just get a silent failed boot.
> 
> I meant that it's broken to use WARN_ON(), because it's a userspace triggerable
> condition.

This and the blacklist case are now two known cases, so yes I'a agree
now. It was not widely known before.

> WARN_ON() is for kernel bugs only.  Of course, if it's a useful
> warning, it can still be left in as pr_warn().

I'll send a patch.

> > > However, request_module() should also
> > > correctly return an error when it fails.  So let's make it return
> > > -ENOENT, which matches the error when the modprobe binary doesn't exist.
> > 
> > This is a user experience change though, and I wouldn't have on my radar
> > who would use this, and expects the old behaviour. Josh, would you by
> > chance?
> > 
> > I'd like this to be more an RFC first so we get vetted parties to
> > review. I take it this and Neil's case are cases we should revisit now,
> > properly document as we didn't before, ensure we don't break anything,
> > and also extend the respective kmod selftests to ensure we don't break
> > these corner cases in the future.
> 
> This patch only affects kernel internals, not the userspace API.

Ah yes, in that case this seems fine with me.

> So I don't see
> why it would be controversial?  I already went through all callers of
> request_module() that check its return value, and they all appear to work better
> with -ENOENT, since they assume that 0 means the module was loaded.

Thanks for doing that, but I note that getting 0 is not assurance
either. The de-facto best practive for the request_module() call is to
do your own in place verifier.

> Incorrectly returning 0 typically causes unnecessary work (checking again
> whether the module's functionality is available) or misleading log messages.

Yes but returning 0 cannot be relied upon today for assuming the module
is loaded. *If* we revisit that decision and want the kernel to do a
generic verifier, then yes, we can get rid of all the caller specific
verfifiers, but not today.

> In
> fact, I can't think of a situation where kernel code would *want* 0 returned in
> this case, as it's ambiguous with the module being successfully loaded.

Unfortunately that's just how the API (to my mind silly) grew out to.

> Sure, I'll check whether it would be possible to add a test for this case in
> lib/test_kmod.c and tools/testing/selftests/kmod/.

Thanks!

  Luis

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

On Wed, Mar 11, 2020 at 06:31:30AM +0000, Luis Chamberlain wrote:
> On Tue, Mar 10, 2020 at 10:26:20PM -0700, Eric Biggers wrote:
> > On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> > > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > > From: Eric Biggers <ebiggers@google.com>
> > > > 
> > > > It's long been possible to disable kernel module autoloading completely
> > > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > > preferable
> > > 
> > > preferable but ... not documented. Or was this documented or recommended
> > > somewhere?
> > > 
> > > > to setting it to a nonexistent file since it avoids the
> > > > overhead of an attempted execve(), avoids potential deadlocks, and
> > > > avoids the call to security_kernel_module_request() and thus on
> > > > SELinux-based systems eliminates the need to write SELinux rules to
> > > > dontaudit module_request.
> > 
> > Not that I know of, though I didn't look too hard.  proc(5) mentions
> > /proc/sys/kernel/modprobe but doesn't mention the empty string case.
> > 
> > In any case, it's been supported for a long time, and it's useful for the
> > reasons I mentioned.
> 
> Sure. I think then its important to document it as such then, or perhaps
> make a kconfig option which sets this to empty and document it on the
> kconfig entry.

I'll send a man-pages patch to document it in proc(5).

Most users, including the one I have in mind, should just be able to run
'echo > /proc/sys/kernel/modprobe' early in the boot process.  So I don't think
the need for a kconfig option to control the default value has been clearly
demonstrated yet.  You're certainly welcome to send a patch for it if you
believe it would be useful, though.

> > So I don't see
> > why it would be controversial?  I already went through all callers of
> > request_module() that check its return value, and they all appear to work better
> > with -ENOENT, since they assume that 0 means the module was loaded.
> 
> Thanks for doing that, but I note that getting 0 is not assurance
> either. The de-facto best practive for the request_module() call is to
> do your own in place verifier.
> 
> > Incorrectly returning 0 typically causes unnecessary work (checking again
> > whether the module's functionality is available) or misleading log messages.
> 
> Yes but returning 0 cannot be relied upon today for assuming the module
> is loaded. *If* we revisit that decision and want the kernel to do a
> generic verifier, then yes, we can get rid of all the caller specific
> verfifiers, but not today.
> 

Sure, I understand all that; I think we're actually on the same page.  Even if
we make the return value of request_module() completely correct, nothing stops
another process from loading or unloading the module immediately afterwards.

However, callers do sometimes use the return value opportunisticly, like to log
an appropriate message or to return early if module loading failed.  Those seem
like relatively appropriate uses.  The thing which you really can't do, which I
didn't see anyone doing, is use 'ret == 0' as a signal to go ahead and run code
that will crash if the module has been unloaded already.

- Eric

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Wed, Mar 11, 2020 at 10:35:45AM -0700, Eric Biggers wrote:
> On Wed, Mar 11, 2020 at 06:31:30AM +0000, Luis Chamberlain wrote:
> > On Tue, Mar 10, 2020 at 10:26:20PM -0700, Eric Biggers wrote:
> > > On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> > > > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > > > From: Eric Biggers <ebiggers@google.com>
> > > > > 
> > > > > It's long been possible to disable kernel module autoloading completely
> > > > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > > > preferable
> > > > 
> > > > preferable but ... not documented. Or was this documented or recommended
> > > > somewhere?
> > > > 
> > > > > to setting it to a nonexistent file since it avoids the
> > > > > overhead of an attempted execve(), avoids potential deadlocks, and
> > > > > avoids the call to security_kernel_module_request() and thus on
> > > > > SELinux-based systems eliminates the need to write SELinux rules to
> > > > > dontaudit module_request.
> > > 
> > > Not that I know of, though I didn't look too hard.  proc(5) mentions
> > > /proc/sys/kernel/modprobe but doesn't mention the empty string case.
> > > 
> > > In any case, it's been supported for a long time, and it's useful for the
> > > reasons I mentioned.
> > 
> > Sure. I think then its important to document it as such then, or perhaps
> > make a kconfig option which sets this to empty and document it on the
> > kconfig entry.
> 
> I'll send a man-pages patch to document it in proc(5).
> 
> Most users, including the one I have in mind, should just be able to run
> 'echo > /proc/sys/kernel/modprobe' early in the boot process.  So I don't think
> the need for a kconfig option to control the default value has been clearly
> demonstrated yet.  You're certainly welcome to send a patch for it if you
> believe it would be useful, though.

When doing a rewrite of some of this code I did wonder who would use
this and clear it out. A kconfig entry would remove any doubt over its
use and would allow one to skip the userspace / early init requirement
to empty it out, therefore actually being safer because you are not
racing against modules being loaded.

Is avoiding the race more suitable for your needs than echo'ing early on boot?

  Luis

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

On Wed, Mar 11, 2020 at 06:00:02PM +0000, Luis Chamberlain wrote:
> On Wed, Mar 11, 2020 at 10:35:45AM -0700, Eric Biggers wrote:
> > On Wed, Mar 11, 2020 at 06:31:30AM +0000, Luis Chamberlain wrote:
> > > On Tue, Mar 10, 2020 at 10:26:20PM -0700, Eric Biggers wrote:
> > > > On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> > > > > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > > > > From: Eric Biggers <ebiggers@google.com>
> > > > > > 
> > > > > > It's long been possible to disable kernel module autoloading completely
> > > > > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > > > > preferable
> > > > > 
> > > > > preferable but ... not documented. Or was this documented or recommended
> > > > > somewhere?
> > > > > 
> > > > > > to setting it to a nonexistent file since it avoids the
> > > > > > overhead of an attempted execve(), avoids potential deadlocks, and
> > > > > > avoids the call to security_kernel_module_request() and thus on
> > > > > > SELinux-based systems eliminates the need to write SELinux rules to
> > > > > > dontaudit module_request.
> > > > 
> > > > Not that I know of, though I didn't look too hard.  proc(5) mentions
> > > > /proc/sys/kernel/modprobe but doesn't mention the empty string case.
> > > > 
> > > > In any case, it's been supported for a long time, and it's useful for the
> > > > reasons I mentioned.
> > > 
> > > Sure. I think then its important to document it as such then, or perhaps
> > > make a kconfig option which sets this to empty and document it on the
> > > kconfig entry.
> > 
> > I'll send a man-pages patch to document it in proc(5).
> > 
> > Most users, including the one I have in mind, should just be able to run
> > 'echo > /proc/sys/kernel/modprobe' early in the boot process.  So I don't think
> > the need for a kconfig option to control the default value has been clearly
> > demonstrated yet.  You're certainly welcome to send a patch for it if you
> > believe it would be useful, though.
> 
> When doing a rewrite of some of this code I did wonder who would use
> this and clear it out. A kconfig entry would remove any doubt over its
> use and would allow one to skip the userspace / early init requirement
> to empty it out, therefore actually being safer because you are not
> racing against modules being loaded.
> 
> Is avoiding the race more suitable for your needs than echo'ing early on boot?
> 

Maybe.  It would avoid the chance of races, but I haven't seen any yet.
Also, our userspace has to support old kernels, so we still need the
'echo > /proc/sys/kernel/modprobe' anyway.  If that turns out to be good enough,
then it makes things easier for everyone.

If setting the default at build time turns out to be needed, then sure in that
case I'll send a patch that adds a kconfig option to do that.  But I'm first
trying to use the existing kernel functionality.

Also, a kconfig option isn't really a substitute for documenting this existing
sysctl.  We still need to document it properly in proc(5) and
Documentation/admin-guide/sysctl/kernel.rst.

- Eric

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Wed, Mar 11, 2020 at 11:21:30AM -0700, Eric Biggers wrote:
> On Wed, Mar 11, 2020 at 06:00:02PM +0000, Luis Chamberlain wrote:
> > On Wed, Mar 11, 2020 at 10:35:45AM -0700, Eric Biggers wrote:
> > > On Wed, Mar 11, 2020 at 06:31:30AM +0000, Luis Chamberlain wrote:
> > > > On Tue, Mar 10, 2020 at 10:26:20PM -0700, Eric Biggers wrote:
> > > > > On Wed, Mar 11, 2020 at 04:32:21AM +0000, Luis Chamberlain wrote:
> > > > > > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > > > > > From: Eric Biggers <ebiggers@google.com>
> > > > > > > 
> > > > > > > It's long been possible to disable kernel module autoloading completely
> > > > > > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > > > > > preferable
> > > > > > 
> > > > > > preferable but ... not documented. Or was this documented or recommended
> > > > > > somewhere?
> > > > > > 
> > > > > > > to setting it to a nonexistent file since it avoids the
> > > > > > > overhead of an attempted execve(), avoids potential deadlocks, and
> > > > > > > avoids the call to security_kernel_module_request() and thus on
> > > > > > > SELinux-based systems eliminates the need to write SELinux rules to
> > > > > > > dontaudit module_request.
> > > > > 
> > > > > Not that I know of, though I didn't look too hard.  proc(5) mentions
> > > > > /proc/sys/kernel/modprobe but doesn't mention the empty string case.
> > > > > 
> > > > > In any case, it's been supported for a long time, and it's useful for the
> > > > > reasons I mentioned.
> > > > 
> > > > Sure. I think then its important to document it as such then, or perhaps
> > > > make a kconfig option which sets this to empty and document it on the
> > > > kconfig entry.
> > > 
> > > I'll send a man-pages patch to document it in proc(5).
> > > 
> > > Most users, including the one I have in mind, should just be able to run
> > > 'echo > /proc/sys/kernel/modprobe' early in the boot process.  So I don't think
> > > the need for a kconfig option to control the default value has been clearly
> > > demonstrated yet.  You're certainly welcome to send a patch for it if you
> > > believe it would be useful, though.
> > 
> > When doing a rewrite of some of this code I did wonder who would use
> > this and clear it out. A kconfig entry would remove any doubt over its
> > use and would allow one to skip the userspace / early init requirement
> > to empty it out, therefore actually being safer because you are not
> > racing against modules being loaded.
> > 
> > Is avoiding the race more suitable for your needs than echo'ing early on boot?
> > 
> 
> Maybe.  It would avoid the chance of races, but I haven't seen any yet.
> Also, our userspace has to support old kernels, so we still need the
> 'echo > /proc/sys/kernel/modprobe' anyway.  If that turns out to be good enough,
> then it makes things easier for everyone.
> 
> If setting the default at build time turns out to be needed, then sure in that
> case I'll send a patch that adds a kconfig option to do that.  But I'm first
> trying to use the existing kernel functionality.
> 
> Also, a kconfig option isn't really a substitute for documenting this existing
> sysctl.  We still need to document it properly in proc(5) and
> Documentation/admin-guide/sysctl/kernel.rst.

Yes, sure. Whatever mechanism you find *suitable* I think you may the
shiny best new user of it, and so given all that is shared now, I'm sure
you will document do what is needed. Whatever guidance you can provide
based on your experience is of huge value to this little corner of the
kernel.

I just wanted to do away with unclear tribal knowledge and ensure we
support / test whatever you do well moving forward.

  Luis

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Josh Triplett]

On March 11, 2020 5:32:21 AM GMT+01:00, Luis Chamberlain <mcgrof@kernel.org> wrote:
>On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote: 
>> However, request_module() should also
>> correctly return an error when it fails.  So let's make it return
>> -ENOENT, which matches the error when the modprobe binary doesn't
>exist.
>
>This is a user experience change though, and I wouldn't have on my
>radar
>who would use this, and expects the old behaviour. Josh, would you by
>chance?

I don't think this affects userspace. But I'd suggest Ben Hutchings (CCed).

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Wed, Mar 11, 2020 at 06:55:30AM +0100, Josh Triplett wrote:
> On March 11, 2020 5:32:21 AM GMT+01:00, Luis Chamberlain <mcgrof@kernel.org> wrote:
> >On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote: 
> >> However, request_module() should also
> >> correctly return an error when it fails.  So let's make it return
> >> -ENOENT, which matches the error when the modprobe binary doesn't
> >exist.
> >
> >This is a user experience change though, and I wouldn't have on my
> >radar
> >who would use this, and expects the old behaviour. Josh, would you by
> >chance?
> 
> I don't think this affects userspace. But I'd suggest Ben Hutchings (CCed).

It doesn't, so yes no verififcation needed. Thanks the quick response though!

  Luis

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Kees Cook]

On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> From: Eric Biggers <ebiggers@google.com>
> 
> It's long been possible to disable kernel module autoloading completely
> by setting /proc/sys/kernel/modprobe to the empty string.  This can be

Hunh. I've never seen that before. :) I've always used;

echo 1 > /proc/sys/kernel/modules_disabled

Regardless,

Reviewed-by: Kees Cook <keescook@chromium.org>

-Kees

> preferable to setting it to a nonexistent file since it avoids the
> overhead of an attempted execve(), avoids potential deadlocks, and
> avoids the call to security_kernel_module_request() and thus on
> SELinux-based systems eliminates the need to write SELinux rules to
> dontaudit module_request.
> 
> However, when module autoloading is disabled in this way,
> request_module() returns 0.  This is broken because callers expect 0 to
> mean that the module was successfully loaded.
> 
> Apparently this was never noticed because this method of disabling
> module autoloading isn't used much, and also most callers don't use the
> return value of request_module() since it's always necessary to check
> whether the module registered its functionality or not anyway.  But
> improperly returning 0 can indeed confuse a few callers, for example
> get_fs_type() in fs/filesystems.c where it causes a WARNING to be hit:
> 
> 	if (!fs && (request_module("fs-%.*s", len, name) == 0)) {
> 		fs = __get_fs_type(name, len);
> 		WARN_ONCE(!fs, "request_module fs-%.*s succeeded, but still no fs?\n", len, name);
> 	}
> 
> This is easily reproduced with:
> 
> 	echo > /proc/sys/kernel/modprobe
> 	mount -t NONEXISTENT none /
> 
> It causes:
> 
> 	request_module fs-NONEXISTENT succeeded, but still no fs?
> 	WARNING: CPU: 1 PID: 1106 at fs/filesystems.c:275 get_fs_type+0xd6/0xf0
> 	[...]
> 
> Arguably this warning is broken and should be removed, since the module
> could have been unloaded already.  However, request_module() should also
> correctly return an error when it fails.  So let's make it return
> -ENOENT, which matches the error when the modprobe binary doesn't exist.
> 
> Cc: stable@vger.kernel.org
> Cc: Alexei Starovoitov <ast@kernel.org>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Cc: Jeff Vander Stoep <jeffv@google.com>
> Cc: Jessica Yu <jeyu@kernel.org>
> Cc: Kees Cook <keescook@chromium.org>
> Cc: Luis Chamberlain <mcgrof@kernel.org>
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> ---
>  kernel/kmod.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/kmod.c b/kernel/kmod.c
> index bc6addd9152b..a2de58de6ab6 100644
> --- a/kernel/kmod.c
> +++ b/kernel/kmod.c
> @@ -120,7 +120,7 @@ static int call_modprobe(char *module_name, int wait)
>   * invoke it.
>   *
>   * If module auto-loading support is disabled then this function
> - * becomes a no-operation.
> + * simply returns -ENOENT.
>   */
>  int __request_module(bool wait, const char *fmt, ...)
>  {
> @@ -137,7 +137,7 @@ int __request_module(bool wait, const char *fmt, ...)
>  	WARN_ON_ONCE(wait && current_is_async());
>  
>  	if (!modprobe_path[0])
> -		return 0;
> +		return -ENOENT;
>  
>  	va_start(args, fmt);
>  	ret = vsnprintf(module_name, MODULE_NAME_LEN, fmt, args);
> -- 
> 2.25.1.481.gfbce0eb801-goog
> 

-- 
Kees Cook

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

On Wed, Mar 11, 2020 at 10:28:07AM -0700, Kees Cook wrote:
> On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > From: Eric Biggers <ebiggers@google.com>
> > 
> > It's long been possible to disable kernel module autoloading completely
> > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> 
> Hunh. I've never seen that before. :) I've always used;
> 
> echo 1 > /proc/sys/kernel/modules_disabled
> 
> Regardless,
> 
> Reviewed-by: Kees Cook <keescook@chromium.org>
> 

modules_disabled is different because it disables *all* module loading, not just
autoloading.

- Eric

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Kees Cook]

On Wed, Mar 11, 2020 at 10:41:34AM -0700, Eric Biggers wrote:
> On Wed, Mar 11, 2020 at 10:28:07AM -0700, Kees Cook wrote:
> > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > From: Eric Biggers <ebiggers@google.com>
> > > 
> > > It's long been possible to disable kernel module autoloading completely
> > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > 
> > Hunh. I've never seen that before. :) I've always used;
> > 
> > echo 1 > /proc/sys/kernel/modules_disabled
> > 
> > Regardless,
> > 
> > Reviewed-by: Kees Cook <keescook@chromium.org>
> > 
> 
> modules_disabled is different because it disables *all* module loading, not just
> autoloading.

Yes, quite true. Some day I'd love to revisit this series to improve
autoloading sanity checking:
https://github.com/KSPP/linux/issues/24

-- 
Kees Cook

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Luis Chamberlain]

On Wed, Mar 11, 2020 at 10:41:34AM -0700, Eric Biggers wrote:
> On Wed, Mar 11, 2020 at 10:28:07AM -0700, Kees Cook wrote:
> > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > From: Eric Biggers <ebiggers@google.com>
> > > 
> > > It's long been possible to disable kernel module autoloading completely
> > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > 
> > Hunh. I've never seen that before. :) I've always used;
> > 
> > echo 1 > /proc/sys/kernel/modules_disabled
> > 
> > Regardless,
> > 
> > Reviewed-by: Kees Cook <keescook@chromium.org>
> > 
> 
> modules_disabled is different because it disables *all* module loading, not just
> autoloading.

Clarifying this on your patch would be useful, otherwise its lost
tribal knowledge.

 LUis

Re: [PATCH] kmod: make request_module() return an error when autoloading is disabled

[Eric Biggers]

On Wed, Mar 11, 2020 at 06:01:07PM +0000, Luis Chamberlain wrote:
> On Wed, Mar 11, 2020 at 10:41:34AM -0700, Eric Biggers wrote:
> > On Wed, Mar 11, 2020 at 10:28:07AM -0700, Kees Cook wrote:
> > > On Tue, Mar 10, 2020 at 03:37:31PM -0700, Eric Biggers wrote:
> > > > From: Eric Biggers <ebiggers@google.com>
> > > > 
> > > > It's long been possible to disable kernel module autoloading completely
> > > > by setting /proc/sys/kernel/modprobe to the empty string.  This can be
> > > 
> > > Hunh. I've never seen that before. :) I've always used;
> > > 
> > > echo 1 > /proc/sys/kernel/modules_disabled
> > > 
> > > Regardless,
> > > 
> > > Reviewed-by: Kees Cook <keescook@chromium.org>
> > > 
> > 
> > modules_disabled is different because it disables *all* module loading, not just
> > autoloading.
> 
> Clarifying this on your patch would be useful, otherwise its lost
> tribal knowledge.

I think it would be more useful to improve the documentation in proc(5) and
Documentation/admin-guide/sysctl/kernel.rst.  People shouldn't have to read
random kernel commit messages to find the documentation.

I'll send out patches for those.

- Eric