| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20150710133912.GI23038@l.oracle.com> Date: Fri, 10 Jul 2015 09:39:12 -0400 From: Konrad Rzeszutek Wilk <konrad.wilk@...cle.com> To: Juergen Gross <jgross@...e.com> Cc: linux-kernel@...r.kernel.org, xen-devel@...ts.xensource.com, david.vrabel@...rix.com, boris.ostrovsky@...cle.com Subject: Re: [Patch V5 00/16] xen: support pv-domains larger than 512GB On Fri, Jul 10, 2015 at 02:47:45PM +0200, Juergen Gross wrote: > Support 64 bit pv-domains with more than 512GB of memory. > > Following test have been done: > - 64 bit dom0 on 8GB machine > - 64 bit dom0 on 1TB machine (resolving p2m/E820-map conflict) > - 32 bit dom0 on 8GB machine > - 64 bit dom0 on 8GB machine with faked kernel/E820-map conflict > - 64 bit dom0 on 8GB machine with faked pgtable/E820-map conflict > - 64 bit dom0 on 8GB machine with faked initrd/E820-map conflict > - 64 bit dom0 on 8GB machine with faked p2m/E820-map conflict > - 64 bit domU (sizes up to 900GB) > - 32 bit domU Woot! Thank you for doing that. I noticed that you used the 'normal' sizes, but nothing as odd as '3145M' or such to try to break things. Not sure if it is worht it - as the test suite that Boris runs does run with some odd sizes to catch this. I only had one comment about the #13 patch, otherwise I think we are OK with this patchset? > > Changes in V5: > - support architectures with undefined KERNEL_PAGE_RO in patch 12 > - correct memblock_reserve() area of kernel page tables on 32 bit > - add some more comments in patch 1 > - some modifications in patch 13 (memblock_reserve for shared_info removed, > for xenstore only necessary if not dom0) > > Changes in V4: > - new patch 13 (add explicit memblock_reserve() calls for special pages) > > Changes in V3: > - rename xen_chk_e820_reserved() to xen_is_e820_reserved() as requested by > David Vrabel > - add __initdata tag to global variables in patch 10 > - move initrd conflict checking after reserving p2m memory (patch 11) > > Changes in V2: > - some clarifications and better explanations in commit messages > - add header changes of include/xen/interface/xen.h (patch 01) > - add wmb() when incrementing p2m_generation (patch 02) > - add new patch 03 (don't build mfn tree if tools don't need it) > - add new patch 06 (split counting of extra memory pages from remapping) > - add new patch 07 (check memory area against e820 map) > - replace early_iounmap() with early_memunmap() (patch 07->patch 08) > - rework patch 09 (check for kernel memory conflicting with memory layout) > - rework patch 10 (check pre-allocated page tables for conflict with memory map) > - combine old patches 08 and 11 into patch 11 > - add new patch 12 (provide early_memremap_ro to establish read-only mapping) > - rework old patch 12 (if p2m list located in to be remapped region delay > remapping) to copy p2m list in case of a conflict (now patch 13) > - correct Kconfig dependency (patch 13->14) > - don't limit dom0 to 512GB (patch 13->14) > - modify parameter parsing to work in very early boot (patch 13->14) > - add new patch 15 to do some cleanup > - remove old patch 05 (simplify xen_set_identity_and_remap() by using global > variables) > - remove old patch 08 (detect pre-allocated memory interfering with e820 map) > > > Juergen Gross (16): > xen: sync with xen headers > xen: save linear p2m list address in shared info structure > xen: don't build mfn tree if tools don't need it > xen: eliminate scalability issues from initial mapping setup > xen: move static e820 map to global scope > xen: split counting of extra memory pages from remapping > xen: check memory area against e820 map > xen: find unused contiguous memory area > xen: check for kernel memory conflicting with memory layout > xen: check pre-allocated page tables for conflict with memory map > xen: check for initrd conflicting with e820 map > mm: provide early_memremap_ro to establish read-only mapping > xen: add explicit memblock_reserve() calls for special pages > xen: move p2m list if conflicting with e820 map > xen: allow more than 512 GB of RAM for 64 bit pv-domains > xen: remove no longer needed p2m.h > > Documentation/kernel-parameters.txt | 7 + > arch/x86/include/asm/xen/interface.h | 96 +++++++- > arch/x86/include/asm/xen/page.h | 8 +- > arch/x86/xen/Kconfig | 20 +- > arch/x86/xen/enlighten.c | 4 +- > arch/x86/xen/mmu.c | 397 ++++++++++++++++++++++++++++++--- > arch/x86/xen/p2m.c | 43 +++- > arch/x86/xen/p2m.h | 15 -- > arch/x86/xen/setup.c | 414 ++++++++++++++++++++++++++--------- > arch/x86/xen/xen-head.S | 2 + > arch/x86/xen/xen-ops.h | 7 + > include/asm-generic/early_ioremap.h | 2 + > include/asm-generic/fixmap.h | 3 + > include/xen/interface/xen.h | 35 +-- > mm/early_ioremap.c | 12 + > 15 files changed, 867 insertions(+), 198 deletions(-) > delete mode 100644 arch/x86/xen/p2m.h > > -- > 2.1.4 > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@...r.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists