[mojo-edk] Serialise windows handles into an "extra header" section.

[mojo-edk] Serialise windows handles into an "extra header" section.

This change introduces an "extra header" section of a channel message
header. On windows, platform handles are place into this section instead
of being appended to the end of a message.

At the same time, make a couple of smaller changes:
1. Allow the number of serialised handles to be less than the number
stated in the constructor.
2. Add a has_mach_ports() helper function.
3. Limit the number of attached handles.

These changes are all prep work for Mach ports support.

BUG=582468
Committed: https://crrev.com/fbebd41187198e7782b3fec9ddf9f155b9df1789
Cr-Commit-Position: refs/heads/master@{#378099}

[Anand Mistry (off Chromium), 2016-02-26 08:53:45]

amistry@chromium.org changed reviewers:
+ rockot@chromium.org

[Ken Rockot(use gerrit already), 2016-02-26 23:07:12]

lgtm with a question: why do we want max handles vs num handles?

[Anand Mistry (off Chromium), 2016-02-27 00:36:22]

Because when forwarding mach ports through the broker, handles need to be
detached from the message and attached to the wrapper message (and the
reason for that is because you can have a mix of fds and mach ports). That
makes the number of handles go from N to 0, and back to N when the message
is unwrapped in the broker process. The easiest way to do this without
extra allocs/memcpy is to just separate the actual number of handles from
the requested number at message construction.

On 27 February 2016 at 10:07, <rockot@chromium.org> wrote:

> lgtm with a question: why do we want max handles vs num handles?
>
> https://codereview.chromium.org/1740953002/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Ken Rockot(use gerrit already), 2016-02-27 00:37:44]

got it, thanks

On Fri, Feb 26, 2016 at 4:36 PM, Anand K. Mistry <amistry@chromium.org>
wrote:

> Because when forwarding mach ports through the broker, handles need to be
> detached from the message and attached to the wrapper message (and the
> reason for that is because you can have a mix of fds and mach ports). That
> makes the number of handles go from N to 0, and back to N when the message
> is unwrapped in the broker process. The easiest way to do this without
> extra allocs/memcpy is to just separate the actual number of handles from
> the requested number at message construction.
>
> On 27 February 2016 at 10:07, <rockot@chromium.org> wrote:
>
>> lgtm with a question: why do we want max handles vs num handles?
>>
>> https://codereview.chromium.org/1740953002/
>>
>
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Anand Mistry (off Chromium), 2016-02-27 02:01:40]

The CQ bit was checked by amistry@chromium.org

[commit-bot: I haz the power, 2016-02-27 02:03:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1740953002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1740953002/40001

[commit-bot: I haz the power, 2016-02-27 03:18:28]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-02-27 03:19:17]

Description was changed from

==========
[mojo-edk] Serialise windows handles into an "extra header" section.

This change introduces an "extra header" section of a channel message
header. On windows, platform handles are place into this section instead
of being appended to the end of a message.

At the same time, make a couple of smaller changes:
1. Allow the number of serialised handles to be less than the number
stated in the constructor.
2. Add a has_mach_ports() helper function.
3. Limit the number of attached handles.

These changes are all prep work for Mach ports support.

BUG=582468
==========

to

==========
[mojo-edk] Serialise windows handles into an "extra header" section.

This change introduces an "extra header" section of a channel message
header. On windows, platform handles are place into this section instead
of being appended to the end of a message.

At the same time, make a couple of smaller changes:
1. Allow the number of serialised handles to be less than the number
stated in the constructor.
2. Add a has_mach_ports() helper function.
3. Limit the number of attached handles.

These changes are all prep work for Mach ports support.

BUG=582468

Committed: https://crrev.com/fbebd41187198e7782b3fec9ddf9f155b9df1789
Cr-Commit-Position: refs/heads/master@{#378099}
==========

[commit-bot: I haz the power, 2016-02-27 03:19:17]

Patchset 3 (id:??) landed as
https://crrev.com/fbebd41187198e7782b3fec9ddf9f155b9df1789
Cr-Commit-Position: refs/heads/master@{#378099}

[Ken Rockot(use gerrit already), 2016-03-01 18:31:09]

A revert of this CL (patchset #3 id:40001) has been created in
https://codereview.chromium.org/1753723003/ by rockot@chromium.org.

The reason for reverting is: Reverting per ARC++ request. Let's reland in a way
that doesn't affect non-OSX POSIX. :|.

[Anand Mistry (off Chromium), 2016-03-01 20:19:20]

On 2016/03/01 18:31:09, Ken Rockot wrote:
> A revert of this CL (patchset #3 id:40001) has been created in
> https://codereview.chromium.org/1753723003/ by mailto:rockot@chromium.org.
> 
> The reason for reverting is: Reverting per ARC++ request. Let's reland in a
way
> that doesn't affect non-OSX POSIX. :|.

I still think this CL is the best way forward and don't think we should be
committing to a stable wire format for the EDK. Maybe one day, but it's entirely
inappropriate at this stage.

[Ken Rockot(use gerrit already), 2016-03-01 20:20:52]

I agree. I am going to make a change ASAP to get the ARC bridge off of
ChannelPosix.

On Tue, Mar 1, 2016 at 12:19 PM, <amistry@chromium.org> wrote:

> On 2016/03/01 18:31:09, Ken Rockot wrote:
> > A revert of this CL (patchset #3 id:40001) has been created in
> > https://codereview.chromium.org/1753723003/ by mailto:
> rockot@chromium.org.
> >
> > The reason for reverting is: Reverting per ARC++ request. Let's reland
> in a
> way
> > that doesn't affect non-OSX POSIX. :|.
>
> I still think this CL is the best way forward and don't think we should be
> committing to a stable wire format for the EDK. Maybe one day, but it's
> entirely
> inappropriate at this stage.
>
> https://codereview.chromium.org/1740953002/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.