Re: [PATCH RFC v9 19/51] x86/sev: Introduce snp leaked pages list
From: Jeremi Piotrowski <hidden>
Date: 2023-08-09 12:46:34
Also in:
kvm, linux-crypto, linux-mm, lkml
On Sun, Jun 11, 2023 at 11:25:27PM -0500, Michael Roth wrote:
quoted hunk ↗ jump to hunk
From: Ashish Kalra <ashish.kalra@amd.com> Pages are unsafe to be released back to the page-allocator, if they have been transitioned to firmware/guest state and can't be reclaimed or transitioned back to hypervisor/shared state. In this case add them to an internal leaked pages list to ensure that they are not freed or touched/accessed to cause fatal page faults. Signed-off-by: Ashish Kalra <ashish.kalra@amd.com> [mdr: relocate to arch/x86/coco/sev/host.c] Signed-off-by: Michael Roth <redacted> --- arch/x86/coco/sev/host.c | 28 ++++++++++++++++++++++++++++ arch/x86/include/asm/sev-host.h | 3 +++ 2 files changed, 31 insertions(+)diff --git a/arch/x86/coco/sev/host.c b/arch/x86/coco/sev/host.c index cd3b4c6a25bc..373e91f5a337 100644 --- a/arch/x86/coco/sev/host.c +++ b/arch/x86/coco/sev/host.c@@ -64,6 +64,12 @@ struct rmpentry { static unsigned long rmptable_start __ro_after_init; static unsigned long rmptable_end __ro_after_init; +/* list of pages which are leaked and cannot be reclaimed */ +static LIST_HEAD(snp_leaked_pages_list); +static DEFINE_SPINLOCK(snp_leaked_pages_list_lock); + +static atomic_long_t snp_nr_leaked_pages = ATOMIC_LONG_INIT(0); + #undef pr_fmt #define pr_fmt(fmt) "SEV-SNP: " fmt@@ -494,3 +500,25 @@ int rmp_make_shared(u64 pfn, enum pg_level level) return rmpupdate(pfn, &val); } EXPORT_SYMBOL_GPL(rmp_make_shared); + +void snp_leak_pages(unsigned long pfn, unsigned int npages) +{ + struct page *page = pfn_to_page(pfn); + + WARN(1, "psc failed, pfn 0x%lx pages %d (marked offline)\n", pfn, npages); + + spin_lock(&snp_leaked_pages_list_lock); + while (npages--) { + /* + * Reuse the page's buddy list for chaining into the leaked + * pages list. This page should not be on a free list currently + * and is also unsafe to be added to a free list. + */ + list_add_tail(&page->buddy_list, &snp_leaked_pages_list); + sev_dump_rmpentry(pfn); + pfn++; + } + spin_unlock(&snp_leaked_pages_list_lock); + atomic_long_inc(&snp_nr_leaked_pages); +} +EXPORT_SYMBOL_GPL(snp_leak_pages);diff --git a/arch/x86/include/asm/sev-host.h b/arch/x86/include/asm/sev-host.h index 753e80d16433..bab3b226777a 100644 --- a/arch/x86/include/asm/sev-host.h +++ b/arch/x86/include/asm/sev-host.h@@ -19,6 +19,8 @@ void sev_dump_rmpentry(u64 pfn); int psmash(u64 pfn); int rmp_make_private(u64 pfn, u64 gpa, enum pg_level level, int asid, bool immutable); int rmp_make_shared(u64 pfn, enum pg_level level); +void snp_leak_pages(unsigned long pfn, unsigned int npages); + #else static inline int snp_lookup_rmpentry(u64 pfn, bool *assigned, int *level) { return 0; } static inline void sev_dump_rmpentry(u64 pfn) {}@@ -29,6 +31,7 @@ static inline int rmp_make_private(u64 pfn, u64 gpa, enum pg_level level, int as return -ENODEV; } static inline int rmp_make_shared(u64 pfn, enum pg_level level) { return -ENODEV; } +void snp_leak_pages(unsigned long pfn, unsigned int npages) {}
This needs to be 'static inline' or the build fails with multiple definition errors. I'm building a guest kernel with CONFIG_KVM_AMD_SEV disabled. Jeremi
#endif #endif -- 2.25.1