lkml.org 
[lkml]   [2019]   [Jan]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] livepatch: core: Return ENOTSUPP instead of ENOSYS
On Tue 2019-01-29 10:50:54, Josh Poimboeuf wrote:
> On Mon, Jan 28, 2019 at 02:49:43PM -0500, Joe Lawrence wrote:
> > On Sun, Jan 27, 2019 at 04:26:30AM +0900, Alice Ferrazzi wrote:
> > > This patch fixes a checkpatch warning:
> > > WARNING: ENOSYS means 'invalid syscall nr' and nothing else
> > >
> > > Signed-off-by: Alice Ferrazzi <alice.ferrazzi@miraclelinux.com>
> > > ---
> > > kernel/livepatch/core.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
> > > index 5b77a7314e01..eea6b94fef89 100644
> > > --- a/kernel/livepatch/core.c
> > > +++ b/kernel/livepatch/core.c
> > > @@ -897,7 +897,7 @@ int klp_register_patch(struct klp_patch *patch)
> > >
> > > if (!klp_have_reliable_stack()) {
> > > pr_err("This architecture doesn't have support for the livepatch consistency model.\n");
> > > - return -ENOSYS;
> > > + return -ENOTSUPP;
> > > }
> > >
> > > return klp_init_patch(patch);
> > > --
> > > 2.19.2
> > >
> >
> > Hi Alice,
> >
> > Patches should be based off the upstream livepatching tree, found here:
> >
> > git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching.git
> >
> > and in this case, the for-next branch, which holds patches that have
> > already been queued up for the next release. This one:
> >
> > 958ef1e39d24 ("livepatch: Simplify API by removing registration step")
> >
> > has moved the code in question from klp_register_patch() to
> > klp_enable_patch().
> >
> >
> > As far as the change itself, I don't have strong opinion about it
> > either way.
> >
> > On the one hand, there is the checkpatch warning and -ENOTSUPP reads
> > more intuitively than -ENOSYS.
> >
> > However, the current pattern seems to be more prevelent in the kernel.
> > I wonder if the checkpatch warning would be better specified for return
> > values that are actually passed back to userspace.
> >
> > Also, klp_register_patch(), now klp_enable_patch(), is exported for
> > module use, though I don't believe anyone (samples / tests / kpatch /
> > kgraft?) is inspecting which error value is returned.
> >
> > I would defer to whichever convention the maintainers prefer here.
>
> Based on the commit description from 91c9afaf97ee ("checkpatch.pl: new
> instances of ENOSYS are errors"), it sounds like there was a decision at
> Kernel Summit to limit ENOSYS to mean "bad syscall" and nothing else.

Hmm, the error code is passed to the syscall, for example:

+ SYSCALL_DEFINE3(init_module
+ load_module()
+ do_init_module()
+ do_one_initcall(mod->init);

I am not sure if we are allowed to return -ENOTSUPP (-524).
It is defined in the internal include/linux/errno.h. There
is the following commnent:

/*
* These should never be seen by user programs...



I tried to find a better alternative and found:

#define EOPNOTSUPP 95 /* Operation not supported on transport endpoint */


There is the following note in man errno:

ENOTSUP Operation not supported (POSIX.1)

EOPNOTSUPP Operation not supported on socket (POSIX.1)
(ENOTSUP and EOPNOTSUPP have the same value
on Linux, but according to POSIX.1 these error
values should be distinct.)

And it looks that -EOPNOTSUPP is used widely in many subsystes (not
only network).

Best Regards,
Petr

\
 
 \ /
  Last update: 2019-01-30 13:42    [W:0.052 / U:5.388 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site