Linux: set the process title (that shows in "ps" etc.) of renderers correctly when using the zygote.

Linux: set the process title (that shows in "ps" etc.) of renderers correctly when using the zygote.
BUG=none
TEST=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=25877

[dank, 2009-09-04 04:24:44]

LGTM but you might want to run it by agl.

[dank, 2009-09-04 04:25:07]

Also you might want to do the BSD port straight off
since they have a syscall for this?

[agl, 2009-09-04 17:27:05]

I don't think this patch should land. The minimal benefit does not justify it.
If, for some reason, people feel strongly about this, then there's a couple of
fixes needed below:

http://codereview.chromium.org/196009/diff/1/2
File base/command_line.cc (right):

http://codereview.chromium.org/196009/diff/1/2#newcode13
Line 13: #include <stdint.h>
In general, stdint shouldn't be needed. basictypes.h should bring it in .

http://codereview.chromium.org/196009/diff/1/2#newcode193
Line 193: if ((reinterpret_cast<uintptr_t>(environ) & ~(PAGE_SIZE - 1)) ==
Holy smokes Batman! Can't you hook in at ChromeMain? This code makes baby Jesus
cry.

http://codereview.chromium.org/196009/diff/1/2#newcode226
Line 226: ~(PAGE_SIZE - 1);
From man getpagesize():

Generally, one uses binaries that are dependent on the  architecture  but  not 
on  the machine model, in order to have a single binary distribution per
architecture.  This means that a user program should not find PAGE_SIZE at
compile time from a header file, but use an  actual system  call,  at least for
those architectures (like sun4) where this dependency exists.

[Mike Mammarella, 2009-09-09 19:45:56]

I've been asked several times by various people how to tell which chrome
processes are renderers, or why there are so many zygotes. And if the number of
people using native process management tools to measure memory use on Windows is
any indication, we'll get a lot of people running ps etc. to look at Chrome's
fancy multi-process model after the Linux beta launch. I think it would be nice
for them to see --type=renderer for renderers, and only one --type=zygote.

I've rearranged the patch so the Chromium code calls an implementation of
setproctitle() which is a BSD API for doing this. I've also commented exactly
how and why it works the way it does: in particular, it seems that the kernel
folks say this is the "official" way to do this.

Do you still think the patch makes juvenile deities cry?

[agl, 2009-09-09 22:14:32]

We have Shift+Esc and about:memory, I still don't see that we want this patch.
However, in current form, it's ok to land if you want to.

http://codereview.chromium.org/196009/diff/7004/7008
File base/setproctitle.c (right):

http://codereview.chromium.org/196009/diff/7004/7008#newcode48
Line 48: static char** main_argv_ = NULL;
globals are usually named with g_

[not_the_right_dank, 2009-09-09 22:21:40]

I think it's worth it, there will be lots of people using ps and top
in spite of our memory tools.

On Wed, Sep 9, 2009 at 3:14 PM, <agl@chromium.org> wrote:
>
> We have Shift+Esc and about:memory, I still don't see that we want this
> patch.
> However, in current form, it's ok to land if you want to.
>
>
> http://codereview.chromium.org/196009/diff/7004/7008
> File base/setproctitle.c (right):
>
> http://codereview.chromium.org/196009/diff/7004/7008#newcode48
> Line 48: static char** main_argv_ = NULL;
> globals are usually named with g_
>
> http://codereview.chromium.org/196009
>