Re: [PATCH 3/5] userfaultfd.2: Add write-protect mode

[Peter Xu <peterx@redhat.com>]

On Mon, Apr 05, 2021 at 11:13:14PM +0200, Michael Kerrisk (man-pages) wrote:
> Hello Alex, Peter,

Hi, Michael,

[...]

> > @@ -144,6 +170,17 @@ single threaded non-cooperative userfaultfd manager implementations.
> >  .\" and limitations remaining in 4.11
> >  .\" Maybe it's worth adding a dedicated sub-section...
> >  .\"
> > +.PP
> > +Since Linux 5.7, userfaultfd is able to do
> > +synchronous page dirty tracking using the new write-protect register mode.
> > +One should check against the feature bit
> > +.B UFFD_FEATURE_PAGEFAULT_FLAG_WP
> > +before using this feature.
> > +Similar to the original userfaultfd missing mode, the write-protect mode will
> > +generate an userfaultfd message when the protected page is written.
> > +The user needs to resolve the page fault by unprotecting the faulted page and
> 
> You use "user-space" elsewhere, but "the user" here. What is the difference?

I wanted to mean the same idea of user-space indeed.

> 
> More generally, What is "the user" in this context? I think you 
> really mean  something like "the trap-handling thread"" or something
> like that? The same statement applies for various pieces below.

Yes, it's just a more general wording since the trap-handling thread in the
userfaultfd world belongs to the user, or the user-space.

> 
> > +kick the faulted thread to continue.
> 
> What does "kick" mean here. This should be explained.

Sorry about that, "kick" is a very frequently used wording for developers but
probably confusing for manpage readers, it means "wake up".

> 
> > +For more information, please refer to "Userfaultfd write-protect mode" section.
> >  .SS Userfaultfd operation
> >  After the userfaultfd object is created with
> >  .BR userfaultfd (),
> > @@ -179,7 +216,7 @@ or
> >  .BR ioctl (2)
> >  operations to resolve the page fault.
> >  .PP
> > -Starting from Linux 4.14, if the application sets the
> > +Since Linux 4.14, if the application sets the
> >  .B UFFD_FEATURE_SIGBUS
> >  feature bit using the
> >  .B UFFDIO_API
> > @@ -219,6 +256,65 @@ userfaultfd can be used only with anonymous private memory mappings.
> >  Since Linux 4.11,
> >  userfaultfd can be also used with hugetlbfs and shared memory mappings.
> >  .\"
> > +.SS Userfaultfd write-protect mode (since 5.7)
> > +Since Linux 5.7, userfaultfd supports write-protect mode.
> > +The user needs to first check availability of this feature using
> > +.B UFFDIO_API
> > +ioctl against the feature bit
> > +.B UFFD_FEATURE_PAGEFAULT_FLAG_WP
> > +before using this feature.
> > +.PP
> > +To register with userfaultfd write-protect mode, the user needs to initiate the
> 
> You use "user-space" elsewhere, but "the user" here. What is the difference?

(I'm skipping same questions)

[...]

> > +.B UFFDIO_REGISTER
> > +with the mode set to
> > +.BR "UFFDIO_REGISTER_MODE_MISSING | UFFDIO_REGISTER_MODE_WP" .
> > +When there is only
> > +.B UFFDIO_REGISTER_MODE_WP
> > +registered, the userspace will
> > +.I not
> > +receive any message when a missing page is written.
> > +Instead, the userspace will only receive a write-protect page fault message
> > +when an existing but write-protected page got written.
> 
> The wording here is a bit off. What does it mean: "an existing but 
> write-protected page"?

I wanted to emphasize the fact that this write-protect message won't be
generated if the page does not exist at all.  That belongs to the missing page
fault even if a memory write triggerred the page fault.

Thanks,

-- 
Peter Xu