stuff, chalk some non-trivial changes to the
windows sandbox at the very least.
-cpu
On Dec 18, 2:49 pm, Adam Barth wrote:
> On Fri, Dec 18, 2009 at 9:47 AM, Charles Reis wrote:
> > Any other examples of browser state that would be tough to restore? How bad
> > would this be f
There is one more issue. Our process singleton detection (in windows)
only works with the full exe path so if you have two chromes one
running from the userprofile and one running from the programfiles dir
then they both can run and try to access the same profile. This is
not a problem if the use
Yes, the MasterPreferences is what we use so far. It is just a json
file in a particular path. I am not sure if we have enabled this for
the linux build.
Afaik the bookmarks has not been done.
On Dec 19, 1:16 am, Peter Kasting wrote:
> On Sat, Dec 19, 2009 at 12:59 AM, PhistucK wrote:
> > Th
I hear you. The issue is to me endemic of singleton/lazyinstance
AtExit dtor -> Singleton A Object dtor --> Singleton B ctor -> AtExit
registration -> bang
Basically the dtors of singletons do too much or the whole scheme
stinks.
--
Chromium Developers mailing list: chromium-dev@googlegroups.
Nice job. Thanks for doing this.
Now the critic:
Not too happy with the thunk generator:
VirtualAlloc(0, sizeof(_stdcallthunk), MEM_COMMIT | MEM_RESERVE,
PAGE_EXECUTE_READWRITE)
But I am too lazy right now to look at what does the VS2008 thunk
logic looks right now. I would expect the page to be
Carlos
- Build PGO instrument again but could not finish PGO optimized with
a 22 Mb training file. Eats 11 GB while linking and slows to a crawl.
- Got a leased machine (in jail) that has 24 GB of ram, installed
everything and managed to build release. Next stop PGO.
Note: I does not feel tha
t 23, 2:06 pm, Peter Kasting wrote:
> On Fri, Oct 23, 2009 at 2:04 PM, cpu wrote:
> > Do we care about this? care as in crafting a test to detect
> > regressions?
>
> How much we care is probably directly proportional to how mu
dotNet ASP pages have a class named HttpBrowserCapabilities that
returns what the name implies. If you go to the url below using chrome
you'll see an echo (in column 3) of what it thinks of your browser:
http://www.on-the-matrix.com/webtools/HttpBrowserCapabilities.aspx
Is there anything you see
+1 on moving spell to the renderers.
We can memory map in the browser and map again the in renderers.
Hopefully read-only.
We eliminate the sync ipc and do not increase the memory usage.
On Oct 22, 2:42 pm, Steve Vandebogart wrote:
> It's been awhile since I looked at this, but the email I was
I am with the others that don't see "move sqlite to another process"
as a natural outcome of these thread.
If using more memory is the concern, another process uses more memory.
sqlite is not crashing *that* much; yes it was the top crasher for a
while but it was a data race
--~--~-~
Did somebody answer Marc-Antoine question?
I don't see us ever releasing pages. I can fathom that we are not
doing that so I must be reading the code wrong.
On Oct 1, 10:11 am, Erik Kay wrote:
> On Thu, Oct 1, 2009 at 8:49 AM, Mike Belshe wrote:
> > see about:tcmalloc (credit to sgk) - which
to do your bidding.
-cpu
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change email options, or unsubscribe:
http://groups.google.com/group/chromium-dev
-~--~~~~--~~--~--~---
fault") or user --user-data-dir to specify the location for a new
profile.
-cpu
ps. There is a new sqlite wrapper but is not being used extensively
yet.
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, chan
I believe brettw new sqlite wrappers removed that. Maybe he has not
landed it or I misunderstood. Let me check.
Of course even if landed I don't know to what revision you synch to.
Also that points to a bug, possibly sqlite db corruption. Do you mind
filling a bug or having one of the users fill
VS2010 has a lot of things that have been re-written, being hit by a
compiler code generation bug is that last thing you want. I'll say we
need to wait a few months after RTM before we move into that version.
But according to all reports it has a lot of good things in store for
us, including bein
The plot thickens, the main history db is also corrupted:
D:\test\corruptdb>sqlite3.exe "zzz\User Data\Default\History"
SQLite version 3.6.17
Enter ".help" for instructions
Enter SQL statements terminated with a ";"
sqlite> PRAGMA integrity_check;
rowid 40017 missing from index visits_time_index
Larson gave me a profile that can consistently crash windows chrome
(beta). I haven't tried loading the profile myself but we have crash
dumps from it.
It crashes in:
0x69db6e5d [chrome.dll - fts2.c:453] getVarint <--- access
violation at 0x0
0x69db6ef0 [chrome.dll -
1. So how much faster is the linking? I mean the release build
linking was the slow one. Debug linking wasn't that bad.
2. net is dependent of base.. so you could not swap base by itself,
right?
3. The native windows resource management will need to change in the
multiple dlls world. ATL has a
On Aug 24, 10:14 am, Evan Martin wrote:
> On Mon, Aug 24, 2009 at 10:08 AM, Jeremy Moskovich wrote:
> > * Due to some technical limitations with the FF libraries, we need to load
> > them in a separate process. It also doesn't seem like a good idea to run
> > code out of an arbitrary library i
If you followed the instructions in chromium (http://dev.chromium.org/
developers/how-tos/build-instructions-windows) recently and chose to
use VS2008 then this will be of interest.
The order was wrong, I just changed it. You need to install in the
following order:
1. VS2008 RTM
2. Windows SDK 6
an easy interim step, could this work?
Has anybody looked into this?
-cpu
WaitForInputIdle:
http://msdn.microsoft.com/en-us/library/ms687022(VS.85).aspx
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change
On Aug 4, 3:36 pm, nakro wrote:
> Ian, i have a lot of respect to you chrome devs, but i could never
> figure why
> you don't just punch holes in the sandbox when Flash or Java or maybe
> even Reader work
>
In general because sandboxing code that you don't have the source code
and can updat
We need a better way to talk about this perf gain. I agree is 12% ops/
second in that particular set of benchmarks. My recollection is that
we removed LFH because it was using too much memory. We need some form
of normalized score based on memory usage. In other words 12% with 25%
more memory usa
What are the results of this experiment?
On Jul 30, 12:15 pm, Huan Ren wrote:
> I just submitted a change (22080) that disables tcmalloc used on
> Windows platform. The plan is keeping it in trunk for 24 hours and
> then reverting it. The intentions are
> - Having another round of performance
If you have a well-known (yet subpar) anti-virus whose name I don't
want to mention. Your build might fail:
general error c101008d: Failed to write the updated manifest to the
resource of file
"..\chrome\Release\sbox_integration_tests.exe"
Solution: tweak your AV if you can.
--~--~-~--~
Jut to be clear, we are happy that you used chromium to create your
proof-of-concept. I hope you take the comments above as constructive
critic.
The order W3C -> webkit is not set on stone. You can also approach
webkit first, or find out who is the lead on CSS for webkit and talk
directly to that
g from
a renderer.
ReceivedBadMessage() current behavior is to terminate the offending
renderer (call BadMessageTerminateProcess())
BadMessageTerminateProcess() has DCHECK()s and LOG() stuff so if you
are calling ReceivedBadMessage() do not put them in your cod
Mike, yes we (I) increased the number of renderer processes for
machines with lots of ram. I think it tops now to 40 processes.
Our previous limit was not based on calculation but because we had
WaitForMultipleObjects(..) which has a 64 objects maximum and we had 2
objects per process so our limi
Yes, the real reason is that there is an ongoing cost of keep that
version working including extra QA cycles for each release. In terms
of supporting a windows version with very few users we should focus
our efforts on Win7.
But you are welcome to keep an external fork. If there is any
consolatio
filed the bug: http://code.google.com/p/chromium/issues/detail?id=14631
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change email options, or unsubscribe:
http://groups.google.com/group/chromium-dev
-~
I spent a ton of time tracking this so for the record:
If you made a change to some files and gcl change works but when you
try to do gcl upload you get this error:
Got error status from
['svn', 'cat', '']
// Copyright (c) 2006-2009 The Chromim authors ..
===
On May 5, 10:44 am, Scott Hess wrote:
> I agree with the earlier argument about not larding startup with
> things like writing new files to id the coming-up Chrome to
> late-coming instances. An alternative might be to acquire a lock to
> protect the profile, and write an id asynchronously aft
On Apr 30, 3:26 pm, Evan Martin wrote:
> On Thu, Apr 30, 2009 at 3:13 PM, Peter Kasting wrote:
> > On Thu, Apr 30, 2009 at 1:50 PM, cpu wrote:
>
> >> Inhttp://src.chromium.org/viewvc/chrome?view=rev&revision=14983I
> >> removed a CoInitialize()/CoUnInitial
Utility process is an amenable idea. We do something like that for
first-run import as well.
Key items, I can think of:
1- Utility process would not display UI (would it?)
2- We can allow a directory to be available for read/write
3- Use IPC for progress / heartbeat
In other words pretty much a
In http://src.chromium.org/viewvc/chrome?view=rev&revision=14983 I
removed a CoInitialize()/CoUnInitialize() pair in the renderer process
of your favorite browser.
This should be the last one of them. COM should now be unusable from
the renderer.
As far as I can see, this was a leftover and we d
ellow chromiunistas if they feel like
increasing the default level one more notch.
This change only affects chrome. Other modules are unit tests and
shell_test are not affected.
-cpu
On Apr 22, 11:52 am, Evan Martin wrote:
> code.google.com is world-writable, so many of the docs there involv
As a hard and fast rule you can consider any crash where we are not
intentionally trying to crash (using __debugbreak(), DebugBreak(),
RaiseException() or CHECK ) as probably exploitable.
If you think a little bit about a crash you might be tempted to think
it is not exploitable, but is easy to g
tion of the executable (the pages marked as read-execute),
they are known at compile time and it would not make sense to
construct them on the fly.
But if you know of a case then that would be very interesting.
>
> Nicolas
>
>
>
> On Thu, Apr 2, 2009 at 1:54 PM, cpu wrote:
>
&
e code base.
... so and what about the manbearpig ? Ah, yes no longer a myth:
http://www.thinkgene.com/scientists-successfully-create-human-bear-pig-chimera/
-cpu
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change email options, or unsubscribe:
http://groups.google.com/group/chromium-dev
-~--~~~~--~~--~--~---
On Apr 2, 11:12 am, Marshall Greenblatt
wrote:
> Hi All,
>
> Is there a current document that explains the multi-process crash
> service/reporting design used by Chrome via crash_service? I see the
> documents available athttp://code.google.com/p/google-breakpad/w/listbut
> they appear a bit d
On Mar 27, 6:39 pm, "Book'em Dano" wrote:
> I'd like to better understand how the interception mechanism works. Is
> it using IAT rewriting?
We use several methods, depending on the task at hand.
Check sandbox_types.h InterceptionType:
- EAT
- Preamble patch (two ways)
- Service call
You c
On windows the exe/dll separation it was meant to simplify upgrade but
in practice the
exe changes frequently enough that we still need to support exe
upgrading in place.
We always wanted to have a do nothing exe that just loads the chrome
dll but that never
happened, reasons: 1) the sandbox 2) b
On Mar 25, 2:52 pm, Mike Reed wrote:
> Indeed, its unhinted to keep it small. That said, we could (with some
> $$) have it hinted, either generically, or with cleartype in mind.
Our font fallback is incomplete, Uniscribe is a no-go due to the
sandbox. This seems something worth exploring.
For
On Mar 22, 8:47 am, empriser wrote:
> There are more than 50,000 patches onhttp://codereview.chromium.org/;
I don't think so. 50k/365 = 140. That means 140 patches per day
including weekends. We are very active but not that active.
> I have some questions about it:
> How many patches have bee
As darin/adam said, the right way is with an ipc message. But I am
curious, what do you want to do?
If it is for debugging purposes you can just start chrome with --no-
sandbox.
If you want to have dialog boxes pop directly controlled by the
renderer, you will be fighting the architecture of chr
11k directories for me on that path. Awesome find.
>> ProjectSection(WebsiteProperties) = preProject
Can't we just delete this section? or would VS put it back? if so
maybe we can put some
nonsense there so VS does not re-create it.
On Feb 19, 3:47 pm, Finnur Thorarinsson wrote:
> Dang. Erik
Are there any implications for sandboxing on the fork vs exec ? I
don't want us to paint ourselves in a corner when we implement the
sandbox.
On Feb 5, 9:57 am, Rahul Kuchhal wrote:
> If file structure on Linux is anywhere like Windows than the shared library
> (chrome.dll on Windows) would be
+1 to string16
I can't make performance or memory saving claims with a straight face
for any. We just don't process enough strings for us to matter.
On Feb 4, 9:57 am, Mike Belshe wrote:
> The big string area is webkit, of course. If webkit were 100% UTF-8
> already, we might take a different
We don't launch renderers using LaunchApp, we use broker_service-
>SpawnTarget(). I guess in other platforms that don't have a sandbox
you can replace that for whatever you want.
You can see BrowserRenderProcessHost::Init() for all the cruft that we
need to launch a renderer, I don't see a good w
as already discussed long time ago.
If somebody asked me that they want to contribute a port of chrome on
Windows UI using MFC, I would say no. I just don't see the cost/
benefit.
Personally, Qt seems now the stronger toolkit, but I really don't have
a clue about linux development.
-cpu
If this is a common scenario, we might need to use another signal.
On Dec 6, 6:25 pm, Peter Kasting <[EMAIL PROTECTED]> wrote:
> On Fri, Dec 5, 2008 at 7:01 PM, cpu <[EMAIL PROTECTED]> wrote:
> > Due to a series of changes in the last month, the app now thinks it
&
On Dec 2, 6:09 pm, Marc-Antoine Ruel <[EMAIL PROTECTED]> wrote:
> I like the idea of fixing test_shell and not trying to convert
> everything at first.
>
> But don't work on 5025, I think it's wasted time. It's a non-essential plugin.
>
..
> There's zero reference to WMIUtil in the code so to f
IsPerUserInstall() in the codebase and ask your
doctor to see if this path is for you.
-cpu
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"Chromium-dev" group.
To post to this group, send email to ch
Ben, look at atlwin.h CWindowImplBaseT< TBase, TWinTraits
>::WindowProc
I believe that OnFinalMessage(window) is called right there if msg ==
WM_NCDESTROY
right about line 3101 on that file.
Does that help you sort out this?
On Nov 23, 3:08 pm, "Ben Goodger (Google)" <[EMAIL PROTECTED]> wrot
;pure
computation').
Mike, do you wan't to cross reference the windows sandbox with the mac
preliminary stuff. I would be good for somebody that lands in one to
go see the other even if the mac is just preliminary thoughts.
-cpu
On Nov 10, 5:28 pm, Darin Fisher <[EMAIL PROTECT
55 matches
Mail list logo