Re: [ros-dev] Debug log problems

[Timo Kreuzer]

Am 06.06.2012 17:42, schrieb Cameron Gutman:
>   3. What happened to the floppy driver?
>  (/srv/buildbot_cmake/full_cmake/build/ntoskrnl/io/iomgr/driver.c:1579)
> '\Driver\FLOPPY' initialization failed, status (0xc000000e)
>  (/srv/buildbot_cmake/full_cmake/build/ntoskrnl/io/iomgr/driver.c:61) Deleting driver
> object '\Driver\FLOPPY'
>  (/srv/buildbot_cmake/full_cmake/build/ntoskrnl/io/iomgr/driver.c:87) HACK: Not unloading
> the driver image due to critical bugs!
>  (/srv/buildbot_cmake/full_cmake/build/ntoskrnl/io/iomgr/driver.c:1961)
> IopInitializeDriver() failed (Status c000000e)
>  (/srv/buildbot_cmake/full_cmake/build/ntoskrnl/mm/ARM3/sysldr.c:935) Leaking driver:
> floppy.sys
>
>  Fix it? Don't enable it in trunk builds if it's buggy and not fixed?
>  That's correct behavior. Failing DriverEntry makes the driver immediately unload.
> This is broken due to some incorrect memory typing in FreeLoader which results in invalid
> loader entries
>  for boot start drivers after stage 1, which is why we currently act very cautiously and
> don't unload to avoid reading invalid memory. This isn't actually a broken case,
> but we don't unload
>  anyway because we have no nice way of knowing whether our LDR entries are valid or not.
Is there a bug report for that? If yes please assign to me.

>  
>
> > 4. These lines repeated 20 times.
> > (/srv/buildbot_cmake/full_cmake/build/dll/ntdll/ldr/ldrapi.c:811) NtCreateSection()
> failed (Status 0xc0000022).
> > Why creating section in LdrpVerifyImageMatchesChecksum() failes now?
> >
> > 5. Win32 stage.
> > (/srv/buildbot_cmake/full_cmake/build/win32ss/user/ntuser/winsta.c:208) err: Invalid
> window station handle 
This is probably due to the fact that certain parts of the
win32k 
initialization are not being done correctly. So it's a real error and 
someone should investigate it.

>  >
> (/srv/buildbot_cmake/full_cmake/build/win32ss/user/ntuser/ntuser.c:161) err: Enter
> NtUserInitialize(0, 00000000, 00000000)
> > (/srv/buildbot_cmake/full_cmake/build/win32ss/user/ntuser/display.c:79) err:
> InitDisplayDriver
> > (\Device\Video0, \Registry\Machine\System\CurrentControlSet\Hardware
> Profiles\Current\System\CurrentControlSet\Services\VBE\Device0);
> >
> > What happens here? It's a normal bootup, everyone should see this. Why is invalid
> window station handle being passed, why do we need NtUserInitialize to produce ERR debug
> print, along with InitDisplayDriver. Are those errors? If not, then why the hell are they
> being printed? If yes, then I don't see any info about actual errors. 
These
should in fact be demoted to TRACEs