Yes, that is weird. Do you see the same issue in February 24 image? Or is this a new issue?
It seems some other module allocates RuntimeServicesData region in ReadyToBootEvent handler. I am not sure if it is introduced by recent HiiDatabaseDxe change. Would you please try to set *PcdHiiOsRuntimeSupport|FALSE*? I want to narrow down the issue. If that is problem, we need consider a better place to construct the table. Thank you Yao Jiewen > -----Original Message----- > From: edk2-devel [mailto:edk2-devel-boun...@lists.01.org] On Behalf Of > Ard Biesheuvel > Sent: Friday, April 1, 2016 8:19 PM > To: Yao, Jiewen <jiewen....@intel.com> > Cc: edk2-devel@lists.01.org <edk2-de...@ml01.01.org>; Gao, Liming > <liming....@intel.com> > Subject: Re: [edk2] [patch V3] MdeModulePkg: Fix Memory Attributes table > type issue > > On 1 April 2016 at 14:16, Ard Biesheuvel <ard.biesheu...@linaro.org> wrote: > > On 1 April 2016 at 09:25, Ard Biesheuvel <ard.biesheu...@linaro.org> > wrote: > >> On 1 April 2016 at 09:24, Ard Biesheuvel <ard.biesheu...@linaro.org> > wrote: > >>> On 31 March 2016 at 15:33, Ard Biesheuvel <ard.biesheu...@linaro.org> > wrote: > >>>> On 31 March 2016 at 15:20, Yao, Jiewen <jiewen....@intel.com> > wrote: > >>>>> Hi Ard > >>>>> Thanks for your log. Yes, it seems new issue, I have never > encountered before. > >>>>> > >>>>> Here is what I found: > >>>>> > >>>>> 1) MemoryAttributeTable is always installed in ReadyToBoot event. > (Sorry, I gave wrong info on EndOfDxe, that is for old properties table) > >>>>> > >>>> > >>>> OK > >>>> > >>>>> 2) In this log, I have seen MemoryAttributeTable is installed 6 times. > >>>>> It > means this system boot 6 boot options. First 5 fail, and last one success. I > am > not clear what device you want to boot, but it seems "Booting EFI Misc > Device" category. It is also weird that RT_Data entry is changed every time. > Does this boot device allocate/free RT_Data? > >>>>> > >>>> > >>>> Good question, I can investigate > >>>> > >>>>> 3) In this log, it shows MemoryAttributeTable is always correct. See > line 1427. I do not know why it is corrupt in efimemmap.log. > >>>>> This memory attribute table is allocated in "BootServicesData" type > memory. Are you sure that system does not corrupt BootServicesData when > efimemmap.log is created? > >>>>> > >>>>> 4) So I suggest we check *when* this table is corrupted. I have > MemoryAttributesDump UEFI APPLICATION (See attachment, password: > 12345678). Maybe can you help run it in UEFI shell to see if > MemoryAttributes table is correct or not? > >>>>> We can use this way to narrow down the corruption - *when* it is > created, or *after* it is created. > >>>>> > >>>> > >>>> Shell> MemoryAttributesDump.efi > >>>> MemoryAttributesTable: > >>>> Version - 0x00000001 > >>>> NumberOfEntries - 0x00000019 > >>>> DescriptorSize - 0x00000030 > >>>> > >>>> Type Start End # Pages > Attributes > >>>> RT_Data 00000000F5B50000-0000000000000000 > 0000000000000030 8000000000004000 > >>>> RT_Code 00000000F5B80000-0000000000000000 > 0000000000000010 8000000000004000 > >>> > >>> Actually, I did find something strange here: these first two regions > >>> are both set to EFI_MEMORY_XP, but the second one is > >>> EfiRuntimeServicesCode > >>> > >>> I am not sure if this is related, but it does look wrong to me. > >>> > >> > >> Never mind, this is mandated by the spec for PE/COFF images. > > > > OK, looking at the debug output again, there is in fact a problem on > > the BIOS side. > > Please refer to the attached debug log. > > > > It shows this region in the memory map > > Available 00000000F5B50000-00000000F5B6FFFF 0000000000000020 > 000000000000000F > > and this region in the memory attributes table > > RT_Data 00000000F5B50000-0000000000000000 0000000000000030 > 8000000000004000 > > This confuses the OS code, since it does not expect the available > region in the memory map, but there is clearly something wrong on the > firmware side. _______________________________________________ edk2-devel mailing list edk2-devel@lists.01.org https://lists.01.org/mailman/listinfo/edk2-devel
