On 05/18/2017 09:39 PM, Dmitry Safonov wrote: > Here are: > - Adjustments to ms API for arch_prctl(ARCH_MAP_VDSO_32). > - Rework of my ms kernel patch for mremapping vdso to vzkernel > (vdso part was rewritten in ~v3.16 so here is a bycicle for older kernel) > - Fix for mmap() bug which resulted in returning pointer over 4GB for > ia32 syscalls (the largest part of the set). > > With those patches + 1 additional patch to vz-criu (u5 branch), > it's possible to C/R ia32 CTs. > I've sent the patch to criu-ml, it's trivial and affects only ia32 C/R, > so soon I'll apply it to the vz7-u5 branch. > > There are two issues known to me: > - zdtm autofs test (it hangs, so no C/R was tested). > - vsyscall page appears on all 32-bit tasks after C/R. > It's emulated, placed over 4Gb, so the only side-effect is > for applications, reading /proc/.../maps and those are not > ready to meet such mapping (such applications not known to me). > > I've tested C/R on centos-6-x86 ia32 container with vzctl, > all looks fine to me. > > Dmitry Safonov (9): > x86/vdso/prctl: fail if vdso present in ARCH_MAP_VDSO_32 > x86/mm: Support mremap()'ing vdso vma > x86/vdso/prctl: Return size of vdso blob for ARCH_MAP_VDSO_32 > ms/x86/mm: Introduce arch_rnd() to compute 32/64 mmap random base > ms/x86/mm: Add task_size parameter to mmap_base() > ms/x86/mm: Introduce mmap_compat_base() for 32-bit mmap() > ms/x86/mm: Make mmap(MAP_32BIT) work correctly > ms/x86/hugetlb: Adjust to the new native/compat mmap bases > x86/mm: Make in_compat_syscall() work during exec > > Kees Cook (2): > ms/x86: standardize mmap_rnd() usage > ms/mm: expose arch_mmap_rnd when available >
Reviewed-by: Andrey Ryabinin <[email protected]> _______________________________________________ Devel mailing list [email protected] https://lists.openvz.org/mailman/listinfo/devel
