[haiku-bugs] Re: [Haiku] #8007: [app_server] fully unresponsive when resizing window of big textfile
- From: "ttcoder" <trac@xxxxxxxxxxxx>
- Date: Mon, 07 Nov 2011 08:59:47 -0000
#8007: [app_server] fully unresponsive when resizing window of big textfile
----------------------------------+----------------------------
Reporter: ttcoder | Owner: axeld
Type: bug | Status: new
Priority: normal | Milestone: R1
Component: Servers/app_server | Version: R1/Development
Resolution: | Keywords:
Blocked By: | Blocking:
Has a Patch: 0 | Platform: All
----------------------------------+----------------------------
Comment (by ttcoder):
At first I was tempted to try it out on my own, but seeing the trouble you
went through and the fact I have no particular skill in kernel-
land/locking/etc to give me hope I'll fare better, I'm very tempted to use
the "let's not duplicate work!" card ;-)
After you upload maybe we could create a "digest" version of the file with
excerpts that seem interesting, for analysis by Ingo ..etc to make their
job easier..
I'm guessing one of the haiku crew will have an "elementary my dear
watson!" moment eventually, if provided with enough hints to narrow down
the list of suspects:
- if a lock is responsible, it would be one common to app_server ''and''
media_server, so maybe that narrows things down to the kernel?
- there might be something specific to single-core systems (this has yet
to be confirmed by someone with a dual/quad core doing the simple
experiment of disabling all but one cores and replicating the bug though..
or to the contrary, reporting that the freeze occurs even with all cores
enabled.. -- oh this reminds me, I'll ping Sean and try to bring him in to
this ticket :-)
If I had more time I'd try to code a minimalist .cpp that reliably
reproduces the freeze, maybe a ten-line program that hammers
FrameResized()... I could end up realizing this does not freeze anything
and it takes an additional ResizeTo() call to freeze the system, which
could hint at the video driver being involved (those have a kernel-land
counterpart to the user-land accelerant after all, probably involving
locks of some sort ?). Or if it didn't, it would hint at the app_server
..etc and so on.
--
Ticket URL: <http://dev.haiku-os.org/ticket/8007#comment:9>
Haiku <http://dev.haiku-os.org>
Haiku - the operating system.
Other related posts:
