[ros-kernel] RE: [ros-cvs] CVS Update: reactos (sections)
James Tabor
jimtabor at adsl-64-217-116-74.dsl.hstntx.swbell.net
Sun Aug 8 05:24:42 CEST 2004
Hi,
This may have fixed paged pool, but I see something else wrong.
Filip Navara wrote:
> Hartmut Birr wrote:
>
>> Hi,
>>
>> it seems that something is wrong in your changes. Compiling ros on ros
>> runs
>> out of non paged pool space. The memory of section object is not freed.
>>
>> - Hartmut
>>
>>
>
> Does this fix patch fix it?
>
> Regards,
> Filip
>
> --- ntoskrnl/mm/section.c 5 Aug 2004 19:59:13 -0000 1.156
> +++ ntoskrnl/mm/section.c 7 Aug 2004 19:40:13 -0000
> @@ -2957,10 +2986,7 @@
> 0,
> NULL,
> SectionHandle);
> - if (!NT_SUCCESS(Status))
> - {
> - ObDereferenceObject(SectionObject);
> - }
> + ObDereferenceObject(SectionObject);
> }
>
> return Status;
This is the end of the compiling the depend files.
Random start:
(KERNEL32:mem/global.c:413) Memory Load: 28
(ex/sysinfo.c:1046) SystemFullMemoryInformation
PID: 1, KernelTime: 18708 DpcTime: 4 PFFree: 0 PFUsed: 65536
MC_CACHE 26433, MC_USER 2876, MC_PPOOL 771, MC_NPPOOL 4612, MmStats.NrFreePages
88168
snip
(KERNEL32:mem/global.c:413) Memory Load: 29
(ex/sysinfo.c:1046) SystemFullMemoryInformation
PID: 1, KernelTime: 19202 DpcTime: 4 PFFree: 0 PFUsed: 65536
MC_CACHE 28024, MC_USER 2585, MC_PPOOL 773, MC_NPPOOL 4736, MmStats.NrFreePages
86742
Stoped
Start compiling taskmgr.
Started new compile:
(KERNEL32:mem/global.c:413) Memory Load: 29
(ex/sysinfo.c:1046) SystemFullMemoryInformation
PID: 1, KernelTime: 22282 DpcTime: 4 PFFree: 0 PFUsed: 65536
MC_CACHE 28130, MC_USER 2427, MC_PPOOL 773, MC_NPPOOL 4747, MmStats.NrFreePages
86783
snip
(KERNEL32:mem/global.c:413) Memory Load: 28
(ex/sysinfo.c:1046) SystemFullMemoryInformation
PID: 1, KernelTime: 22340 DpcTime: 5 PFFree: 0 PFUsed: 65536
MC_CACHE 27467, MC_USER 2143, MC_PPOOL 773, MC_NPPOOL 4723, MmStats.NrFreePages
87754
Compile ended:
Problem is the memory load should have dropped to 9 or 10 percent between the stop
and the start of the new compiling. Looks like we are not freeing MC_USER and
MC_CACHE properly.
Thanks,
James
More information about the Ros-kernel
mailing list