Hi,
On Tue, Jul 07, 2009 at 01:21:09PM +0800, Da Zheng wrote:
> olafbuddenha...@gmx.net wrote:
>> Well, AIUI, there are only two ways how a receive right could be used:
>> either mach_port_insert_right() (or mach_port_extract_right()) is
>> invoked, in which case we can look at the parameters, so
Hi,
On Mon, Jul 06, 2009 at 02:26:10PM +0800, Da Zheng wrote:
> olafbuddenha...@gmx.net wrote:
> Since you think it's not proper to wrap the first thread at the place
> where trace exec_startup_get_info() is called, I now wrap the first
> thread of a task when the first RPC request comes.
Actual
olafbuddenha...@gmx.net wrote:
mach_reply_port() cannot be traced by the rpctrace at all since it's a
system trap.
I see... So there is no way to divert it? That seems strange :-(
Anyways, while this means that we won't see a port created by
mach_reply_port() immediately, we still can
Hi,
olafbuddenha...@gmx.net wrote:
I'm not convinced to use ihash everywhere by that ihash is a library in
Hurd and its code is maintained.
I still think we need to use the hash table where it is needed, use the
linked list where it is needed, etc.
I guess you don't like the dynamic array:-).
Hi,
On Mon, Jun 22, 2009 at 05:22:48PM +0800, Da Zheng wrote:
> olafbuddenha...@gmx.net wrote:
>>> Though Hurd provides ihash, we cannot use it as a dynamic array or
>>> linked list. They are very common structure. We should have a general
>>> implementation and include them in some library such
Hi,
On Mon, Jun 15, 2009 at 04:04:43PM +0800, Zheng Da wrote:
> Unfortunately, git in my Hurd doesn't support --patience option, so
> the patch might be still messy in some places.
1.6.3.1 is available from the debian-ports repository
-antrik-
olafbuddenha...@gmx.net wrote:
I think it is still useful even when rpctrace is tracing a single task
program. rpctrace can know whether a send right is to a traced task or
not and thus, don't need to treat mach_port_insert_right() specially
and can handle the case below properly.
Well,
Hi,
On Mon, Jun 08, 2009 at 09:27:29PM +0800, Da Zheng wrote:
> olafbuddenha...@gmx.net wrote:
>> On Sun, May 31, 2009 at 11:32:25AM +0800, Zheng Da wrote:
>>> 1. tracing the source and destination of the RPC requests/replies.
>>> It is required when we move the port rights and it also allows
>>>
Hi,
Below is the updated patch. The most important update is that I
separate the messy traced_info into 3 structures. There are other
modifications according to antrik's comments.
Unfortunately, git in my Hurd doesn't support --patience option, so
the patch might be still messy in some places.
The
Hi,
On Sun, May 31, 2009 at 11:32:25AM +0800, Zheng Da wrote:
> I rewrite rpctrace so that it can trace multitask programs.
I wonder, is this inherently a single large change, or would it somehow
be possible to split it into several distinct patches, which could be
applied individually?...
> 1.
Hi,
I rewrite rpctrace so that it can trace multitask programs.
There are two major modifications:
1. tracing the source and destination of the RPC requests/replies. It
is required when we move the port rights and it also allows rpctrace
to show the caller of RPCs (I think it is a nice feature).
11 matches
Mail list logo
