Re: [Xen-devel] [PATCH for-4.12 RFC] xen/console: Handle NUL character in buffer sent via CONSOLEIO_write

[Julien Grall <julien.grall@xxxxxxx>]

Hi Juergen,

On 3/5/19 2:12 PM, Juergen Gross wrote:
> On 05/03/2019 15:08, Julien Grall wrote:
> > Hi Juergen,
> >
> > On 3/5/19 12:57 PM, Juergen Gross wrote:
> > > On 27/02/2019 11:45, Julien Grall wrote:
> > > > (+ Juergen Gross as RM)
> > > >
> > > > I forgot to CC Juergen for this.
> > > >
> > > > On 2/26/19 11:03 PM, Julien Grall wrote:
> > > > > After upgrading Debian to Buster, I started noticing console mangling
> > > > > when using zsh. This is happenning because output sent by zsh to the
> > > > > console may contain NUL character in the middle of the buffer.
> > > > >
> > > > > Linux is sending the buffer as it is to Xen console via
> > > > > CONSOLEIO_write.
> > > > > However, the implementation in Xen considers NUL character is used to
> > > > > terminate the buffer and therefore will ignore anything after it.
> > > > >
> > > > > The actual documentation of CONSOLEIO_write is pretty limited. From the
> > > > > declaration, the hypercall takes a buffer and size. So this could lead
> > > > > to think the NUL character is allowed in the middle of the buffer.
> > > > >
> > > > > This patch updates the console API to pass the size along the buffer
> > > > > down so we can remove the reliance on buffer terminating by a NUL
> > > > > character.
> > > > >
> > > > > Signed-off-by: Julien Grall <julien.grall@xxxxxxx>
> > > The risk for a regression is too high this late in the 4.12 release
> > > process IMO.
> > This code path is fairly well tested (console are used pretty much all
> > the times). So a regression would be quickly noticed.
> Only if you test all the guests out in the wild.
The buffer is bounded to 'nr'. So the worst things that can happen is 
you print more characters than you wanted. CONSOLEIO_write is using a 
semantics very similar to write() and we didn't document what happen 
when encountering a NUL character. So, I highly doubt anyone relies on 
the current behavior.
Thinking a bit more, from what Ian wrote [1], the issue maybe wider than 
zsh. So maybe we want to write a band-aid patch at least helping the 
most common case (i.e losing all characters after the first NUL character).
The band-aid patch should be contained to just the hypercall. Would that 
be more suitable for you?
> > This patch has the advantage to allow upgrade to a newer Debian without
> > loosing part of your prompt on zsh. I am not sure whether the problem is
> > the same with other Distros.
> >
> > > My plan is to have only one further RC before branching off 4.12,
> > > so please let us shift this patch to 4.13.
> > I understand. It is possible to workaround the problem at least with
> > zsh. So a release note in Xen (and maybe Debian) should do the job.
> Could you please post the needed information for the 4.12 RN?
How about:

"
While the hypercall CONSOLEIO_write looks analogous to the POSIX call 
write, it will only print character up to the first NUL character if any 
in the buffer. This may result to loss of characters for any application 
using directly the POSIX call write.
The issue has been confirmed some zsh version (such as in Debian 
Buster). Where the prompt is mangled, this could be avoided by adding 
'setopt single_line_zle' in your .zshrc.
"

Cheers,

[1] https://lists.xen.org/archives/html/xen-devel/2019-03/msg00082.html

--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel