Re: [PATCHv2 1/3] x86, ptdump: Add section for EFI runtime services

From: Mathias Krause
Date: Tue Oct 07 2014 - 13:08:05 EST

On Tue, Oct 07, 2014 at 05:01:32PM +0200, Borislav Petkov wrote:
> On Fri, Oct 03, 2014 at 02:47:07PM +0100, Matt Fleming wrote:
> > Looks OK to me. Borislav?
>
> It needs more work AFAICT because with it, espfix area gets cut off
> prematurely:
>

I don't think so. See below...

> ...
> [ 0.134611] ---[ Vmemmap ]---
> [ 0.135003] 0xffffea0000000000-0xffffea0002000000 32M RW PSE GLB NX pmd
> [ 0.136743] 0xffffea0002000000-0xffffea0040000000 992M pmd
> [ 0.138091] 0xffffea0040000000-0xffffea8000000000 511G pud
> [ 0.139611] 0xffffea8000000000-0xffffff0000000000 20992G pgd
> [ 0.140610] ---[ ESPfix Area ]---
> [ 0.141003] 0xffffff0000000000-0xffffff8000000000 512G pgd
> [ 0.142614] 0xffffff8000000000-0xffffffef00000000 444G pud
> [ 0.144088] ---[ EFI Runtime Services ]---
> [ 0.144722] 0xffffffef00000000-0xfffffffec0000000 63G pud
> [ 0.146090] 0xfffffffec0000000-0xfffffffefbe00000 958M pmd
> [ 0.147613] 0xfffffffefbe00000-0xfffffffefbfe0000 1920K pte
> [ 0.149003] 0xfffffffefbfe0000-0xfffffffefc000000 128K RW x pte
> [ 0.150484] 0xfffffffefc000000-0xfffffffefc065000 404K pte
> [ 0.151612] 0xfffffffefc065000-0xfffffffefc200000 1644K RW x pte
> [ 0.153285] 0xfffffffefc200000-0xfffffffefc400000 2M RW PSE x pmd
> [ 0.154721] 0xfffffffefc400000-0xfffffffefc5e0000 1920K RW x pte
> ...
>
> and I think we want to see something more from the espfix area (this is
> what we have now):
>
> [ 0.138086] ---[ ESPfix Area ]---
> [ 0.138590] 0xffffff0000000000-0xffffff8000000000 512G pgd
> [ 0.140099] 0xffffff8000000000-0xfffffffec0000000 507G pud
> [ 0.141444] 0xfffffffec0000000-0xfffffffefbe00000 958M pmd
> [ 0.142597] 0xfffffffefbe00000-0xfffffffefbfe0000 1920K pte
> [ 0.144086] 0xfffffffefbfe0000-0xfffffffefc000000 128K RW x pte
> [ 0.145545] 0xfffffffefc000000-0xfffffffefc065000 404K pte
> [ 0.146597] 0xfffffffefc065000-0xfffffffefc200000 1644K RW x pte
> [ 0.148346] 0xfffffffefc200000-0xfffffffefc400000 2M RW PSE x pmd
> [ 0.149776] 0xfffffffefc400000-0xfffffffefc5e0000 1920K RW x pte
> [ 0.151347] 0xfffffffefc5e0000-0xfffffffefc631000 324K pte
> [ 0.152593] 0xfffffffefc631000-0xfffffffefc655000 144K RW x pte
> [ 0.154143] 0xfffffffefc655000-0xfffffffefc801000 1712K pte
> [ 0.155437] 0xfffffffefc801000-0xfffffffefc831000 192K RW x pte
> [ 0.157004] 0xfffffffefc831000-0xfffffffefc881000 320K pte
> [ 0.158088] 0xfffffffefc881000-0xfffffffefca01000 1536K RW x pte
> [ 0.159712] 0xfffffffefca01000-0xfffffffefcb34000 1228K pte
> [ 0.161117] ... 36 entries skipped ...

What you can see here are actually the EFI runtime service mappings, not
the ESP fix area. Check the addresses and compare them. You should find
similarities ;) And, in fact, the EFI mappings are incomplete in the
second dump, i.e. the vanilla kernel one, because of the enforced limit
for the ESP fix area.

So, in your examples are actually *no* ESP fix area mappings as those
would be r/o. In fact, I think, the above dumps are the result of a
CONFIG_EFI_PGT_DUMP enabled kernel that dumps the page table after
setting up the EFI mappings. There are no ESP fix mappings in this dump
because those are only set up after the EFI runtime service mappings.
See the following code in init/main:

#ifdef CONFIG_X86
if (efi_enabled(EFI_RUNTIME_SERVICES))
efi_enter_virtual_mode();
#endif
#ifdef CONFIG_X86_ESPFIX64
/* Should be run before the first non-init thread is created */
init_espfix_bsp();
#endif

To get a more complete view of the mappings, have a look at the debugfs
file /sys/kernel/debug/kernel_page_tables.

For v3.17 I get (notice the missing EFI mappings):
...
---[ Vmemmap ]---
0xffffea0000000000-0xffffff0000000000 21T pgd
---[ ESPfix Area ]---
0xffffff0000000000-0xffffff3b00000000 236G pud
0xffffff3b00000000-0xffffff3b0000e000 56K pte
0xffffff3b0000e000-0xffffff3b0000f000 4K ro GLB NX pte
0xffffff3b0000f000-0xffffff3b0001e000 60K pte
0xffffff3b0001e000-0xffffff3b0001f000 4K ro GLB NX pte
0xffffff3b0001f000-0xffffff3b0002e000 60K pte
0xffffff3b0002e000-0xffffff3b0002f000 4K ro GLB NX pte
0xffffff3b0002f000-0xffffff3b0003e000 60K pte
0xffffff3b0003e000-0xffffff3b0003f000 4K ro GLB NX pte
0xffffff3b0003f000-0xffffff3b0004e000 60K pte
0xffffff3b0004e000-0xffffff3b0004f000 4K ro GLB NX pte
0xffffff3b0004f000-0xffffff3b0005e000 60K pte
0xffffff3b0005e000-0xffffff3b0005f000 4K ro GLB NX pte
0xffffff3b0005f000-0xffffff3b0006e000 60K pte
0xffffff3b0006e000-0xffffff3b0006f000 4K ro GLB NX pte
0xffffff3b0006f000-0xffffff3b0007e000 60K pte
... 131165 entries skipped ...
---[ High Kernel Mapping ]---
0xffffffff80000000-0xffffffff81000000 16M pmd
...

For v3.17 plus this patch I get this:
...
---[ Vmemmap ]---
0xffffea0000000000-0xffffff0000000000 21T pgd
---[ ESPfix Area ]---
0xffffff0000000000-0xffffff5600000000 344G pud
0xffffff5600000000-0xffffff5600005000 20K pte
0xffffff5600005000-0xffffff5600006000 4K ro GLB NX pte
0xffffff5600006000-0xffffff5600015000 60K pte
0xffffff5600015000-0xffffff5600016000 4K ro GLB NX pte
0xffffff5600016000-0xffffff5600025000 60K pte
0xffffff5600025000-0xffffff5600026000 4K ro GLB NX pte
0xffffff5600026000-0xffffff5600035000 60K pte
0xffffff5600035000-0xffffff5600036000 4K ro GLB NX pte
0xffffff5600036000-0xffffff5600045000 60K pte
0xffffff5600045000-0xffffff5600046000 4K ro GLB NX pte
0xffffff5600046000-0xffffff5600055000 60K pte
0xffffff5600055000-0xffffff5600056000 4K ro GLB NX pte
0xffffff5600056000-0xffffff5600065000 60K pte
0xffffff5600065000-0xffffff5600066000 4K ro GLB NX pte
0xffffff5600066000-0xffffff5600075000 60K pte
... 131059 entries skipped ...
---[ EFI Runtime Services ]---
0xffffffef00000000-0xfffffffec0000000 63G pud
0xfffffffec0000000-0xfffffffef8800000 904M pmd
0xfffffffef8800000-0xfffffffef89d0000 1856K pte
0xfffffffef89d0000-0xfffffffef8a00000 192K RW x pte
0xfffffffef8a00000-0xfffffffef8a75000 468K pte
0xfffffffef8a75000-0xfffffffef8c00000 1580K RW x pte
0xfffffffef8c00000-0xfffffffef8e00000 2M RW PSE x pmd
0xfffffffef8e00000-0xfffffffef8fd0000 1856K RW x pte
0xfffffffef8fd0000-0xfffffffef9041000 452K pte
0xfffffffef9041000-0xfffffffef9065000 144K RW x pte
0xfffffffef9065000-0xfffffffef9211000 1712K pte
0xfffffffef9211000-0xfffffffef9241000 192K RW x pte
0xfffffffef9241000-0xfffffffef9291000 320K pte
0xfffffffef9291000-0xfffffffef9411000 1536K RW x pte
0xfffffffef9411000-0xfffffffef9529000 1120K pte
0xfffffffef9529000-0xfffffffef9600000 860K RW x pte
0xfffffffef9600000-0xfffffffefa400000 14M RW PSE x pmd
0xfffffffefa400000-0xfffffffefa491000 580K RW x pte
0xfffffffefa491000-0xfffffffefa528000 604K pte
0xfffffffefa528000-0xfffffffefa529000 4K RW x pte
0xfffffffefa529000-0xfffffffefa708000 1916K pte
0xfffffffefa708000-0xfffffffefa728000 128K RW x pte
0xfffffffefa728000-0xfffffffefa907000 1916K pte
0xfffffffefa907000-0xfffffffefa908000 4K RW x pte
0xfffffffefa908000-0xfffffffefab06000 2040K pte
0xfffffffefab06000-0xfffffffefab07000 4K RW x pte
0xfffffffefab07000-0xfffffffefad05000 2040K pte
0xfffffffefad05000-0xfffffffefad06000 4K RW x pte
0xfffffffefad06000-0xfffffffefae07000 1028K pte
0xfffffffefae07000-0xfffffffefaf05000 1016K RW x pte
0xfffffffefaf05000-0xfffffffefb005000 1M pte
0xfffffffefb005000-0xfffffffefb007000 8K RW x pte
0xfffffffefb007000-0xfffffffefb1ea000 1932K pte
0xfffffffefb1ea000-0xfffffffefb205000 108K RW x pte
0xfffffffefb205000-0xfffffffefb3e1000 1904K pte
0xfffffffefb3e1000-0xfffffffefb3ea000 36K RW x pte
0xfffffffefb3ea000-0xfffffffefb5cf000 1940K pte
0xfffffffefb5cf000-0xfffffffefb600000 196K RW x pte
0xfffffffefb600000-0xfffffffefb800000 2M RW PSE x pmd
0xfffffffefb800000-0xfffffffefb9e1000 1924K RW x pte
0xfffffffefb9e1000-0xfffffffefbb26000 1300K pte
0xfffffffefbb26000-0xfffffffefbbcf000 676K RW x pte
0xfffffffefbbcf000-0xfffffffefbc80000 708K pte
0xfffffffefbc80000-0xfffffffefbd26000 664K RW x pte
0xfffffffefbd26000-0xfffffffefbe7d000 1372K pte
0xfffffffefbe7d000-0xfffffffefbe80000 12K RW x pte
0xfffffffefbe80000-0xfffffffefc05e000 1912K pte
0xfffffffefc05e000-0xfffffffefc07d000 124K RW x pte
0xfffffffefc07d000-0xfffffffefc237000 1768K pte
0xfffffffefc237000-0xfffffffefc25e000 156K RW x pte
0xfffffffefc25e000-0xfffffffefc434000 1880K pte
0xfffffffefc434000-0xfffffffefc437000 12K RW x pte
0xfffffffefc437000-0xfffffffefc62e000 2012K pte
0xfffffffefc62e000-0xfffffffefc634000 24K RW x pte
0xfffffffefc634000-0xfffffffefc82c000 2016K pte
0xfffffffefc634000-0xfffffffefc82c000 2016K pte
0xfffffffefc82c000-0xfffffffefc82e000 8K RW x pte
0xfffffffefc82e000-0xfffffffefca2a000 2032K pte
0xfffffffefca2a000-0xfffffffefca2c000 8K RW x pte
0xfffffffefca2c000-0xfffffffefcc28000 2032K pte
0xfffffffefcc28000-0xfffffffefcc2a000 8K RW x pte
0xfffffffefcc2a000-0xfffffffefce15000 1964K pte
0xfffffffefce15000-0xfffffffefce28000 76K RW x pte
0xfffffffefce28000-0xfffffffefd012000 1960K pte
0xfffffffefd012000-0xfffffffefd015000 12K RW x pte
0xfffffffefd015000-0xfffffffefd20e000 2020K pte
0xfffffffefd20e000-0xfffffffefd212000 16K RW x pte
0xfffffffefd212000-0xfffffffefd40d000 2028K pte
0xfffffffefd40d000-0xfffffffefd40e000 4K RW x pte
0xfffffffefd40e000-0xfffffffefd5e9000 1900K pte
0xfffffffefd5e9000-0xfffffffefd60d000 144K RW x pte
0xfffffffefd60d000-0xfffffffefd7e7000 1896K pte
0xfffffffefd7e7000-0xfffffffefd7e9000 8K RW x pte
0xfffffffefd7e9000-0xfffffffefd9e0000 2012K pte
0xfffffffefd9e0000-0xfffffffefd9e7000 28K RW x pte
0xfffffffefd9e7000-0xfffffffefdbdf000 2016K pte
0xfffffffefdbdf000-0xfffffffefdbe0000 4K RW x pte
0xfffffffefdbe0000-0xfffffffefddce000 1976K pte
0xfffffffefddce000-0xfffffffefdddf000 68K RW x pte
0xfffffffefdddf000-0xfffffffefdfcd000 1976K pte
0xfffffffefdfcd000-0xfffffffefdfce000 4K RW x pte
0xfffffffefdfce000-0xfffffffefe1af000 1924K pte
0xfffffffefe1af000-0xfffffffefe1cd000 120K RW x pte
0xfffffffefe1cd000-0xfffffffefe3ae000 1924K pte
0xfffffffefe3ae000-0xfffffffefe3af000 4K RW x pte
0xfffffffefe3af000-0xfffffffefe5a8000 2020K pte
0xfffffffefe5a8000-0xfffffffefe5ae000 24K RW x pte
0xfffffffefe5ae000-0xfffffffefe7a5000 2012K pte
0xfffffffefe7a5000-0xfffffffefe7a8000 12K RW x pte
0xfffffffefe7a8000-0xfffffffefe96d000 1812K pte
0xfffffffefe96d000-0xfffffffefe9a5000 224K RW x pte
0xfffffffefe9a5000-0xfffffffefeb69000 1808K pte
0xfffffffefeb69000-0xfffffffefeb6d000 16K RW x pte
0xfffffffefeb6d000-0xfffffffefed65000 2016K pte
0xfffffffefed65000-0xfffffffefed69000 16K RW x pte
0xfffffffefed69000-0xfffffffefef5f000 2008K pte
0xfffffffefef5f000-0xfffffffefef65000 24K RW x pte
0xfffffffefef65000-0xfffffffeff0dc000 1500K pte
0xfffffffeff0dc000-0xfffffffeff15f000 524K RW x pte
0xfffffffeff15f000-0xfffffffeff261000 1032K pte
0xfffffffeff261000-0xfffffffeff2dc000 492K RW x pte
0xfffffffeff2dc000-0xfffffffeff45f000 1548K pte
0xfffffffeff45f000-0xfffffffeff460000 4K RW x pte
0xfffffffeff460000-0xfffffffeff600000 1664K pte
0xfffffffeff600000-0xfffffffeff620000 128K RW x pte
0xfffffffeff620000-0xfffffffeff800000 1920K pte
0xfffffffeff800000-0xffffffff00000000 8M RW PSE x pmd
0xffffffff00000000-0xffffffff80000000 2G pud
---[ High Kernel Mapping ]---
...

The ESP fix area is trimmed, as expected. The EFI runtime service area
is, beside being rather lengthy, complete. Also, as expected.

>
> But yeah, this issue needs to be addressed one way or the other as the
> espfix dump skips the runtime services.
>
> And frankly, I don't see where we're setting that ->max_lines thing but
> it sounds like a promising thing to use. :)

It's the third parameter in the address_markers[] array in
arch/x86/mm/dump_pagetables.c:

# ifdef CONFIG_X86_ESPFIX64
{ ESPFIX_BASE_ADDR, "ESPfix Area", 16 },
# endif

So, it's 16 entries for the ESP fix area. If it's set to 0, as it is for
all the other entries, no limitation applies.


Thanks,
Mathias

>
> Thanks.
>
> --
> Regards/Gruss,
> Boris.
> --
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/