Copied from the archive because I will receive the mail in some hours:
>hbirr at svn.reactos.com wrote:
>>- Guard the copying to the IOSB.
>>- Do the main processing on success or if previous STATUS_PENDING was
returned.
>> Do not use some of the IRP and FO flags at this point.
>>
>>
>Ok, I was going to complain about this, but I realize that it must be
>used since defintely some parts of ROS don't work properly with those
>flags set.
It isn't possible to use the flags, because some functions are always
synchronous independently which flags are set. Ntoskrnl may set the
flags correctly, but a driver may possible not.
>>- Set all results before signaling the events.
>>- Signal the FO event previous the user event.
>>- Made the code a little bit shorter.
>>
>>
>>
>I like your changes and have no complaints, except that the signaling
>semantics of the File/User events should be different in the failure
>case (my "else" path). You've completely removed that path and thus
>changed the logic.
If the driver has returned STATUS_PENDING and completes the irp later
with an error, both events must be signaled, because two different
threads may waiting on this events. The IOSB must be also set. The apc
must be also deliver (I've tested this with the apc sample and some
modifications on windows). There is no difference between completing the
irp with and without an error if the driver has returned STATUS_PENDING.
- Hartmut
sedwards(a)svn.reactos.com wrote:
>remove trailing whitespace at end of lines
>
> *[truncated at 1000 lines; 37354 more skipped]*
Steven,
This patch is unacceptably large, and your commit message is way too
short to explain all the changes. Please revert the patch immediately
and work on a branch, submitting small patches at a time so that other
developers can review them. Also, formatting patches are prohibited, so
I don't think your patches can be accepted at all, even if in a branch.
Best regards,
Alex Ionescu
<If anyone did not notice the sarcasm, please shoot yourselves. Great
work Steven; I hoped you used an automated tool!>
hbirr(a)svn.reactos.com wrote:
> - initialize a user profile before loading syssetup.dll.
> - this makes it possible to install ros over an existing ros.
>
>
>
> Updated files:
> trunk/reactos/subsys/system/setup/makefile
> trunk/reactos/subsys/system/setup/setup.c
>
Harmut ,
Does this apply if Ros is started from a freeldr diskette ?
Regards
Gerard
Hi,
There are currently some files in ntoskrnl/rtl, lib/rtl, and /lib/crt
which are duplicated. Stuff like vsprintf, sprintf, etc.
I think they should be massaged into lib/string so that all the modules
can share the code instead of duplicating it THREE times(!).
Best regards,
Alex Ionescu
--- J M <open.proyects(a)gmail.com> a écrit:
> Hi list,
>
> I am having "problems" with debugging support. I have my build
> environment in Windows. I set up gcc/g++/nasm ok and i'm building last
> SVNs without any problem.
>
> I'd like to debug some code and i get problems here. I test vmware
> with the following results ...
>
> VMWare (config => KDB=1, DBG=1)
> ---------------------------------------
> * DEBUGPORT=COM1 BAUDRATE=115200 => CRASH, FREEZE!!!
> * DEBUGPORT=FILE => REBOOT
im getting the same problem with DEBUGPORT=FILE on real hardware.
no clue found at the moment.
>
> Can we debug ReactOS with VMWare ?
>
> How can i update the ReactOS files in a QEMU's .img file ?
mount it with winimage on windows,
and with the loopback under linux.
mount -o lo qemufile.img /mnt/ros_c
Kind regards,
Usurp (aka Sylvain Petreolle)
humans are like computers,
yesterday the BIOS was all
- today its just a word
Hi list,
I am having "problems" with debugging support. I have my build
environment in Windows. I set up gcc/g++/nasm ok and i'm building last
SVNs without any problem.
I'd like to debug some code and i get problems here. I test vmware
with the following results ...
VMWare (config => KDB=1, DBG=1)
---------------------------------------
* DEBUGPORT=COM1 BAUDRATE=115200 => CRASH, FREEZE!!!
* DEBUGPORT=FILE => REBOOT
Can we debug ReactOS with VMWare ?
How can i update the ReactOS files in a QEMU's .img file ?
Thanks in advance!
Hi,
since some time, I'm not able to install ros from the bootcd over an
existing ros. The first stage setup works perfect. The second boot hangs
after loading most of the drivers. The last line on the debug port comes
from framebuf.dll:
...
DriverBase for \??\E:\ReactOS\system32\win32k.sys: 9d62b000
DriverBase for \??\E:\ReactOS\system32\freetype.dll: 9d803000
DriverBase for \SystemRoot\System32\kbdgr.dll: 9d78f000
(ntuser/desktop.c:574) IntShellHookNotify: No desktop!
(ntuser/desktop.c:574) IntShellHookNotify: No desktop!
(ntuser/desktop.c:574) IntShellHookNotify: No desktop!
DriverBase for \SystemRoot\System32\framebuf.DLL: 9dfab000
Has anyone the same problem? The install works again, if I delete all
registry files from the existing installation. Usually I use the 'make
install' command to install ros over the network on my test machines.
- Hartmut
Hi,
I find the problem is caused by msys or by msys's make.exe, the command
line:
$(CABMAN) /C bootdata/packages/reactos.dff /L $(BOOTCD_DIR)/reactos /I
Will be expanded like this:
$(CABMAN) c:/ bootdata/packages/reactos.dff c:/msys/1.0/L
(BOOTCD_DIR)/reactos i:/
I don't know why. And when I use "mingw32-make bootcd", everything is ok.
And if we use cabman.exe in msys, the command line should like this:
$(CABMAN) //C bootdata/packages/reactos.dff //L $(BOOTCD_DIR)/reactos //I
It goes well.
-----邮件原件-----
发件人: ros-dev-bounces(a)reactos.com [mailto:ros-dev-bounces@reactos.com] 代
表 Hartmut Birr
发送时间: 2005年5月5日 19:32
收件人: ReactOS Development List
主题: Re: [ros-dev] about rules.mak, a small bug fixed
Your patch breaks compiling on windows without msys. I've commit a
different fix which will solve your problem. There does exist one
problem, it isn't possible to build the bootcd. There is somewhere a
'c:/..' on the path, which isn't correct within a simulated posix system
like msys.
- Hartmut