KVM: s390: A bunch of gmap-related fixes
Claudio Imbrenda
maintainerv4 cover KVM: s390: A bunch of gmap-related fixes
Fix handling of _PAGE_UNUSED pte bit. The bit was lingering around longer than it should have, causing issues. This is yet a different approach from the previous version.
Fix some cmma dirty tracking functions.
Also fix a few other small issues in gmap.
v3->v4:
* Improve/fix descriptions of patches 1 and 3
* Fix handle_{sske,pfmf} under memory pressure
* Fix cmma dirty tracking and its selftest
v2->v3:
* Fix _PAGE_UNUSED in yet a different way
* Fix a potential overflow in dat_peek_cmma()
* Some minor gmap fixes that are not actual bugs yet
v1->v2:
* Completely different approach, entirely new patch
Claudio Imbrenda (7):
s390/mm: Fix handling of _PAGE_UNUSED pte bit
KVM: s390: Fix dat_peek_cmma() overflow
KVM: s390: Do not set special large pages dirty
KVM: s390: Fix code typo in gmap_protect_asce_top_level()
KVM: s390: Fix handle_{sske,pfmf} under memory pressure
KVM: s390: cmma: Fix cmma dirty tracking
KVM: s390: selftests: Fix cmma selftest
arch/s390/include/asm/pgtable.h | 4 ++-- arch/s390/kvm/dat.c | 5 ++++- arch/s390/kvm/gmap.c | 18 ++++++++++++++---- arch/s390/kvm/kvm-s390.c | 5 +---- arch/s390/kvm/priv.c | 8 ++++++-- arch/s390/mm/gmap_helpers.c | 3 ++- tools/testing/selftests/kvm/s390/cmma_test.c | 3 +++ 7 files changed, 32 insertions(+), 14 deletions(-)
Signature · 1 line
2.54.0
Claudio Imbrenda
maintainerPATCH v4 1/7 s390/mm: Fix handling of _PAGE_UNUSED pte bit
The _PAGE_UNUSED softbit should not really be lying around. Its sole purpose is to signal to try_to_unmap_one() and try_to_migrate_one() that the page can be discarded instead of being moved / swapped.
KVM has no way to know why a page is being unmapped, so it sets the bit on userspace ptes corresponding to unused guest pages every time they get unmapped. KVM has no reasonable way to clear the bit once the page is in use again.
While set_ptes() checks and clears the bit, other paths that set new ptes did not. This led to used pages being thrown out as if they were unused, causing guest corruption.
Fix the issue by clearing the _PAGE_UNUSED bit for present ptes in set_pte(), i.e. whenever a present pte is getting set. The check in set_ptes() is then redundant and can be removed.
Also fix gmap_helper_try_set_pte_unused() to only set the bit if the pte is present; the _PAGE_UNUSED bit is only defined for present ptes and thus should not be set for non-present ptes.
--- arch/s390/include/asm/pgtable.h | 4 ++-- arch/s390/mm/gmap_helpers.c | 3 ++- 2 files changed, 4 insertions(+), 3 deletions(-)
arch/s390/include/asm/pgtable.h +2 -2
@@ -980,6 +980,8 @@ static inline void set_pmd(pmd_t *pmdp, pmd_t pmd)static inline void set_pte (pte_t *ptep,pte_t pte) { +if (pte_present (pte)) + pte =clear_pte_bit (pte,__pgprot (_PAGE_UNUSED));WRITE_ONCE (*ptep, pte); }
@@ -1332,8 +1334,6 @@ pgprot_t pgprot_writecombine(pgprot_t prot);static inline void set_ptes (struct mm_struct *mm,unsigned long addr,pte_t *ptep,pte_t entry,unsigned int nr) { -if (pte_present (entry)) - entry =clear_pte_bit (entry,__pgprot (_PAGE_UNUSED));page_table_check_ptes_set (mm, addr, ptep, entry, nr);for (;;) {set_pte (ptep, entry);
arch/s390/mm/gmap_helpers.c +2 -1
@@ -181,7 +181,8 @@ void gmap_helper_try_set_pte_unused(struct mm_struct *mm, unsigned long vmaddr)if (IS_ERR_OR_NULL (ptep))return ; -__atomic64_or (_PAGE_UNUSED, (long *)ptep); +if (pte_present (*ptep)) +__atomic64_or (_PAGE_UNUSED, (long *)ptep);pte_unmap_unlock (ptep, ptl); }EXPORT_SYMBOL_GPL (gmap_helper_try_set_pte_unused);
Signature · 1 line
2.54.0
Claudio Imbrenda
maintainerPATCH v4 2/7 KVM: s390: Fix dat_peek_cmma() overflow
In reply to Claudio ImbrendaIf userspace passes a start address that is out of bounds, _dat_walk_gfn_range() will fail with -EFAULT, but state.end will not be touched and will stay 0. This will cause *count to underflow and report a very high number, and the function will end up erroneously reporting success.
Fix by only setting *count if the end address is not smaller than the starting address. This way invalid starting addresses will correctly return -EFAULT and *count will correctly indicate that no values have been returned.
--- arch/s390/kvm/dat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/s390/kvm/dat.c +1 -1
@@ -1209,7 +1209,7 @@ int dat_peek_cmma(gfn_t start, union asce asce, unsigned int *count, u8 *values)int rc; rc =_dat_walk_gfn_range (start, start + *count, asce, &ops,DAT_WALK_DEFAULT , &state); - *count = state.end - start; + *count = state.end >= start ? state.end - start :0 ;/* Return success if at least one value was saved, otherwise an error. */ return (rc == -EFAULT && *count >0 ) ?0 : rc; }
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 2/7] KVM: s390: Fix dat_peek_cmma() overflow
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Pre-existing issues:
- [Critical] Uninitialized kernel memory leak in kvm_s390_get_cmma_bits() via vmalloc array.
- [High] kvm_s390_set_cmma_bits() silently ignores memory allocation failures. --
commit 198324f8297a62f137ba34c4a9aa74e25bb30e54
Author: Claudio Imbrenda <[email protected]>
KVM: s390: Fix dat_peek_cmma() overflow
This commit fixes an underflow in dat_peek_cmma() when userspace passes an out-of-bounds start address. By enforcing that the count is set to zero if the end address is smaller than the start address, it correctly propagates the -EFAULT error instead of erroneously reporting success.
arch/s390/kvm/dat.c int dat_peek_cmma(gfn_t start, union asce asce, unsigned int *count, u8 *values) · line 1213
Quoted 11 lines
arch/s390/kvm/dat.c +1
-1@@ -1209,7 +1209,7 @@ int dat_peek_cmma(gfn_t start, union asce asce, unsigned int *count, u8 *values)int rc; rc =_dat_walk_gfn_range (start, start + *count, asce, &ops,DAT_WALK_DEFAULT , &state); - *count = state.end - start; + *count = state.end >= start ? state.end - start :0 ;/* Return success if at least one value was saved, otherwise an error. */
[Severity: Critical]
This isn't a bug introduced by this patch, but does the CMMA retrieval path leak uninitialized kernel memory to userspace?
In kvm_s390_get_cmma_bits(), the values array is allocated with vmalloc():
arch/s390/kvm/kvm-s390.c:kvm_s390_get_cmma_bits() {
...
values = vmalloc(args->count);
if (!values)
...
}During the page table walk, if __dat_peek_cmma_crste() encounters an invalid CRSTE, it advances the boundary but leaves the corresponding state.values unpopulated:
arch/s390/kvm/dat.c:__dat_peek_cmma_crste() {
...
if (crstep->h.i)
state->end = min(walk->end, next);
...
}Similarly, __dat_get_cmma_pte() skips clean pages:
arch/s390/kvm/dat.c:__dat_get_cmma_pte() {
...
if (!READ_ONCE(*pgste_of(ptep)).cmma_d)
return 0;
...
}Because vmalloc() does not zero-initialize memory, could these gaps in the array leak kernel heap memory when the array is unconditionally copied back to userspace using the advanced state.end count?
[Severity: High]
This is a pre-existing issue, but does kvm_s390_set_cmma_bits() silently ignore memory allocation failures?
If vmalloc() fails to allocate the bits array, the code jumps to the out label:
arch/s390/kvm/kvm-s390.c:kvm_s390_set_cmma_bits() {
...
bits = vmalloc(array_size(sizeof(*bits), args->count));
if (!bits)
goto out;
...
}Since the return variable is initialized to 0 and not updated to a negative error code like -ENOMEM before this jump, will this return 0 to userspace and silently drop the CMMA bits, potentially leading to guest memory attribute corruption during migration?
Quoted 2 lines
arch/s390/kvm/dat.c +0
-0return (rc == -EFAULT && *count >0 ) ?0 : rc; }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
Claudio Imbrenda
maintainerPATCH v4 3/7 KVM: s390: Do not set special large pages dirty
In reply to Claudio ImbrendaSpecial pages / folios should not be set dirty. This also applies to large pages.
Add a missing check in gmap_clear_young_crste() to prevent setting the large page dirty if it is a special page.
--- arch/s390/kvm/gmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/s390/kvm/gmap.c +1 -1
@@ -327,7 +327,7 @@ static long gmap_clear_young_crste(union crste *crstep, gfn_t gfn, gfn_t end, st new.h.i =1 ; new.s.fc1.y =0 ; new.s.fc1.prefix_notif =0 ; -if (new.s.fc1.d || !new.h.p) +if ((new.s.fc1.d || !new.h.p) && !new.s.fc1.s)folio_set_dirty (phys_to_folio (crste_origin_large (crste))); new.s.fc1.d =0 ; new.h.p =1 ;
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 3/7] KVM: s390: Do not set special large pages dirty
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Pre-existing issues:
- [Critical] Missing Special Page Validations in PV / Unmap Paths
- [Critical] Unsafe SSKE Execution on Special Pages
--
commit 2d60ecc18379805e67b3a6ad94ddb8516f4d4167
Author: Claudio Imbrenda <[email protected]>
KVM: s390: Do not set special large pages dirty
This commit prevents setting special large pages dirty. It adds a missing check in gmap_clear_young_crste() to ensure the large page is not a special page before setting it dirty.
arch/s390/kvm/gmap.c static long gmap_clear_young_crste(union crste *crstep, gfn_t gfn, gfn_t end, st · line 331
Quoted 10 lines
arch/s390/kvm/gmap.c +1
-1@@ -327,7 +327,7 @@ static long gmap_clear_young_crste(union crste *crstep, gfn_t gfn, gfn_t end, st new.h.i =1 ; new.s.fc1.y =0 ; new.s.fc1.prefix_notif =0 ; -if (new.s.fc1.d || !new.h.p) +if ((new.s.fc1.d || !new.h.p) && !new.s.fc1.s)folio_set_dirty (phys_to_folio (crste_origin_large (crste)));
[Severity: Critical]
This is a pre-existing issue, but I noticed there are other places where special page validation seems to be missing in PV and unmap paths.
If a malicious userspace maps device memory (VM_PFNMAP) to a PV guest, special shadow page table entries are created. When KVM unmaps these pages, _gmap_unmap_crste() calls phys_to_folio() on the physical frame number:
arch/s390/kvm/gmap.c:_gmap_unmap_crste() {
...
if (old.s.fc1.pr && test_bit(GMAP_FLAG_EXPORT_ON_UNMAP, &priv->gmap->flags))
folio = phys_to_folio(crste_origin_large(old));
...
}Could this result in a garbage pointer if the entry is a special page, leading to a host panic when passed to uv_convert_from_secure_folio() which immediately calls folio_get()?
A similar pattern appears in _gmap_unmap_pte() without checking ptep->s.s:
arch/s390/kvm/gmap.c:_gmap_unmap_pte() {
...
if (ptep->s.pr && test_bit(GMAP_FLAG_EXPORT_ON_UNMAP, &priv->gmap->flags))
folio = pfn_folio(ptep->h.pfra);
...
}And in _destroy_pages_crste():
arch/s390/kvm/gmap.c:_destroy_pages_crste() {
...
for ( ; cur < end; cur += PAGE_SIZE)
__kvm_s390_pv_destroy_page(phys_to_page(cur));
...
}As well as _destroy_pages_pte() and _kvm_s390_pv_make_secure() in arch/s390/kvm/pv.c. Do these paths also need special bit checks to prevent panics from malicious userspace or guests?
[Severity: Critical]
This is also a pre-existing issue, but does KVM_S390_RESET_SKEYS safely handle special pages?
When userspace triggers the KVM_S390_RESET_SKEYS ioctl, dat_reset_skeys_pte() and dat_reset_skeys_crste() walk the gmap and invoke page_set_storage_key() without checking if the page is special. For example:
arch/s390/kvm/dat.c:dat_reset_skeys_pte() {
...
if (ptep->s.pr)
page_set_storage_key(pte_origin(*ptep), PAGE_DEFAULT_KEY, 1);
...
}arch/s390/kvm/dat.c:dat_reset_skeys_crste() {
...
for ( ; addr < end; addr += PAGE_SIZE)
page_set_storage_key(addr, PAGE_DEFAULT_KEY, 1);
...
}If page_set_storage_key() executes the hardware sske instruction on a physical address not backed by valid system RAM, could this cause an s390 Addressing Exception and panic the host?
Quoted 2 lines
arch/s390/kvm/gmap.c +0
-0new.s.fc1.d =0 ; new.h.p =1 ;
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3
Claudio Imbrenda
maintainerPATCH v4 4/7 KVM: s390: Fix code typo in gmap_protect_asce_top_level()
In reply to Claudio ImbrendaThe correct length to pass to kvm_s390_get_guest_pages() is asce.tl + 1, not asce.dt + 1. It was a typo, which, due to fortuitous circumstances, did not cause bugs. It should nonetheless be fixed.
--- arch/s390/kvm/gmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/s390/kvm/gmap.c +1 -1
@@ -1262,7 +1262,7 @@ static int gmap_protect_asce_top_level(struct kvm_s390_mmu_cache *mc, struct gma/* Pairs with the smp_wmb() in kvm_mmu_invalidate_end(). */ smp_rmb (); - rc =kvm_s390_get_guest_pages (sg->kvm, context.f, asce.rsto, asce.dt +1 , false); + rc =kvm_s390_get_guest_pages (sg->kvm, context.f, asce.rsto, asce.tl +1 , false);if (rc >0 ) rc = -EFAULT ;if (!rc)
Signature · 1 line
2.54.0
Claudio Imbrenda
maintainerPATCH v4 5/7 KVM: s390: Fix handle_{sske,pfmf} under memory pressure
In reply to Claudio ImbrendaUnder heavy memory pressure, handle_sske() and handle_pfmf() might cause an endless loop if the mmu cache runs empty, the atomic allocations fail, and the top-up function also fails. While quite unlikely, that scenario is not impossible.
Fix the issue by not ignoring the return value of kvm_s390_mmu_cache_topup(), and appropriately returning an error code in case of failure.
--- arch/s390/kvm/priv.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
arch/s390/kvm/priv.c +6 -2
@@ -366,7 +366,9 @@ static int handle_sske(struct kvm_vcpu *vcpu)if (rc >1 )return kvm_s390_inject_program_int (vcpu,PGM_ADDRESSING );if (rc == -ENOMEM ) { -kvm_s390_mmu_cache_topup (vcpu->arch.mc); + rc =kvm_s390_mmu_cache_topup (vcpu->arch.mc); +if (rc) +return rc;continue ; }if (rc <0 )
@@ -1122,7 +1124,9 @@ static int handle_pfmf(struct kvm_vcpu *vcpu)if (rc >1 )return kvm_s390_inject_program_int (vcpu, rc);if (rc == -ENOMEM ) { -kvm_s390_mmu_cache_topup (vcpu->arch.mc); + rc =kvm_s390_mmu_cache_topup (vcpu->arch.mc); +if (rc) +return rc;continue ; }if (rc <0 )
Signature · 1 line
2.54.0
Re: [PATCH v4 5/7] KVM: s390: Fix handle_{sske,pfmf} under memory pressure
Am 19.06.26 um 17:51 schrieb Claudio Imbrenda:
Quoted 11 lines
Under heavy memory pressure,
handle_sske()andhandle_pfmf()might cause an endless loop if the mmu cache runs empty, the atomic allocations fail, and the top-up function also fails. While quite unlikely, that scenario is not impossible.Fix the issue by not ignoring the return value of
kvm_s390_mmu_cache_topup(), and appropriately returning an error code in case of failure.Fixes e38c884df921 ("KVM: s390: Switch to new gmap")Signed-off-by Claudio Imbrenda <[email protected]>
Quoted 30 lines
--- arch/s390/kvm/priv.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)arch/s390/kvm/priv.c +6
-2@@ -366,7 +366,9 @@ static int handle_sske(struct kvm_vcpu *vcpu)if (rc >1 )return kvm_s390_inject_program_int (vcpu,PGM_ADDRESSING );if (rc == -ENOMEM ) { -kvm_s390_mmu_cache_topup (vcpu->arch.mc); + rc =kvm_s390_mmu_cache_topup (vcpu->arch.mc); +if (rc) +return rc;continue ; }if (rc <0 )@@ -1122,7 +1124,9 @@ static int handle_pfmf(struct kvm_vcpu *vcpu)if (rc >1 )return kvm_s390_inject_program_int (vcpu, rc);if (rc == -ENOMEM ) { -kvm_s390_mmu_cache_topup (vcpu->arch.mc); + rc =kvm_s390_mmu_cache_topup (vcpu->arch.mc); +if (rc) +return rc;continue ; }if (rc <0 )
Claudio Imbrenda
maintainerPATCH v4 6/7 KVM: s390: cmma: Fix cmma dirty tracking
In reply to Claudio ImbrendaIt is possible that some guest memory areas have not been touched yet when starting migration mode, and thus have no ptes allocated. Only existing and allocated ptes should count toward the total of dirty cmma entries.
When starting migration mode, count how many pages actually have a pte (and PGSTE), instead of blindly counting the number of pages in all memslots. Avoid double-counting pages whose cmma information has been updated concurrently.
Also fix dat_get_cmma() to properly wrap around if the first attempt reached the end of guest memory without finding cmma-dirty pages.
--- arch/s390/kvm/dat.c | 3 +++ arch/s390/kvm/gmap.c | 14 ++++++++++++-- arch/s390/kvm/kvm-s390.c | 5 +---- 3 files changed, 16 insertions(+), 6 deletions(-)
arch/s390/kvm/dat.c +3 -0
@@ -1253,6 +1253,9 @@ int dat_get_cmma(union asce asce, gfn_t *start, unsigned int *count, u8 *values, };_dat_walk_gfn_range (*start,asce_end (asce), asce, &ops,DAT_WALK_IGN_HOLES , &state); +/* If no dirty pages were found, wrap around and continue searching */ +if (*start && state.start ==-1 ) +_dat_walk_gfn_range (0 , *start, asce, &ops,DAT_WALK_IGN_HOLES , &state);if (state.start ==-1 ) { *count =0 ;
arch/s390/kvm/gmap.c +12 -2
@@ -1075,7 +1075,16 @@ int gmap_protect_rmap(struct kvm_s390_mmu_cache *mc, struct gmap *sg, gfn_t p_gfstatic long __set_cmma_dirty_pte (union pte *ptep,gfn_t gfn,gfn_t next,struct dat_walk *walk) { -__atomic64_or (PGSTE_CMMA_D_BIT , &pgste_of (ptep)->val); +union pgste pgste; + + pgste =pgste_get_lock (ptep); +/* Avoid double-counting when concurrent updates happen */ +if (!pgste.cmma_d) { + pgste.cmma_d =1 ; +atomic64_inc (walk->priv); + } +pgste_set_unlock (ptep, pgste); +if (need_resched ())return next;return 0 ;
@@ -1089,7 +1098,8 @@ void gmap_set_cmma_all_dirty(struct gmap *gmap)do {scoped_guard (read_lock, &gmap->kvm->mmu_lock) gfn = _dat_walk_gfn_range(gfn, asce_end(gmap->asce), gmap->asce, &ops, -DAT_WALK_IGN_HOLES ,NULL ); +DAT_WALK_IGN_HOLES , + &gmap->kvm->arch.cmma_dirty_pages);cond_resched (); }while (gfn); }
arch/s390/kvm/kvm-s390.c +1 -4
@@ -1191,7 +1191,6 @@ static int kvm_s390_vm_start_migration(struct kvm *kvm) {struct kvm_memory_slot *ms;struct kvm_memslots *slots; -unsigned long ram_pages =0 ;int bkt;/* migration mode already enabled */
@@ -1208,12 +1207,10 @@ static int kvm_s390_vm_start_migration(struct kvm *kvm)kvm_for_each_memslot (ms, bkt, slots) {if (!ms->dirty_bitmap)return -EINVAL ; - ram_pages += ms->npages; } + kvm->arch.migration_mode =1 ;/* mark all the pages as dirty */ gmap_set_cmma_all_dirty (kvm->arch.gmap); -atomic64_set (&kvm->arch.cmma_dirty_pages, ram_pages); - kvm->arch.migration_mode =1 ;kvm_s390_sync_request_broadcast (kvm,KVM_REQ_START_MIGRATION );return 0 ; }
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 6/7] KVM: s390: cmma: Fix cmma dirty tracking
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
New issues:
- [High] The cmma_dirty_pages counter permanently leaks and becomes artificially inflated across VM CMMA resets or memslot removals, leading to an infinite polling loop in userspace during the final migration phase.
Pre-existing issues:
- [High] kvm_s390_get_cmma_bits leaks uninitialized kernel memory to userspace because the values buffer is allocated with vmalloc and gaps corresponding to non-dirty pages are never initialized. --
commit ae0c1471cfdb2824a403c917222645798560f3c1
Author: Claudio Imbrenda <[email protected]>
KVM: s390: cmma: Fix cmma dirty tracking
It is possible that some guest memory areas have not been touched yet when starting migration mode, and thus have no ptes allocated. Only existing and allocated ptes should count toward the total of dirty cmma entries.
arch/s390/kvm/dat.c int dat_get_cmma(union asce asce, gfn_t *start, unsigned int *count, u8 *values, · line 1255
Quoted 8 lines
arch/s390/kvm/dat.c +0
-0@@ -1253,6 +1253,9 @@ int dat_get_cmma(union asce asce, gfn_t *start, unsigned int *count, u8 *values, };_dat_walk_gfn_range (*start,asce_end (asce), asce, &ops,DAT_WALK_IGN_HOLES , &state);
[Severity: High]
This is a pre-existing issue, but does kvm_s390_get_cmma_bits() leak uninitialized kernel memory to userspace?
In kvm_s390_get_cmma_bits(), the values array is allocated using vmalloc() which does not clear the memory:
kvm_s390_get_cmma_bits() {
...
values = vmalloc(args->count);
if (!values)
return -ENOMEM;
...
}When dat_get_cmma() populates this array via __dat_get_cmma_pte(), it iterates over the guest's page tables and only writes to indices corresponding to pages where cmma_d == 1.
For pages where cmma_d == 0, the callback simply returns and skips them, leaving those intermediate array indices untouched:
arch/s390/kvm/dat.c:__dat_get_cmma_pte() {
...
if (!READ_ONCE(*pgste_of(ptep)).cmma_d)
return 0;
...
}The total count returned is calculated as state.end - state.start, encompassing the entire span from the first to the last dirty page found, including the skipped uninitialized gaps.
This array is then copied to userspace via copy_to_user(), leaking uninitialized kernel memory.
Quoted 6 lines
arch/s390/kvm/dat.c +3
-0+/* If no dirty pages were found, wrap around and continue searching */ +if (*start && state.start ==-1 ) +_dat_walk_gfn_range (0 , *start, asce, &ops,DAT_WALK_IGN_HOLES , &state);if (state.start ==-1 ) { *count =0 ;
[ ... ]
Quoted 13 lines
arch/s390/kvm/kvm-s390.c +0
-1@@ -1191,7 +1191,6 @@ static int kvm_s390_vm_start_migration(struct kvm *kvm) {struct kvm_memory_slot *ms;struct kvm_memslots *slots; -unsigned long ram_pages =0 ;int bkt;/* migration mode already enabled */
arch/s390/kvm/kvm-s390.c static int kvm_s390_vm_start_migration(struct kvm *kvm) · line 1215
Quoted 9 lines
arch/s390/kvm/kvm-s390.c +1
-2@@ -1208,12 +1207,10 @@ static int kvm_s390_vm_start_migration(struct kvm *kvm)kvm_for_each_memslot (ms, bkt, slots) {if (!ms->dirty_bitmap)return -EINVAL ; - ram_pages += ms->npages; } + kvm->arch.migration_mode =1 ;/* mark all the pages as dirty */ gmap_set_cmma_all_dirty (kvm->arch.gmap); -atomic64_set (&kvm->arch.cmma_dirty_pages, ram_pages);
[Severity: High]
Does removing this reset of cmma_dirty_pages cause the counter to permanently leak and become artificially inflated?
The cmma_dirty_pages counter tracks the number of pages with the cmma_d bit set.
However, when CMMA states are cleared (e.g., via the KVM_S390_VM_MEM_CLR_CMMA ioctl) or when memory is unmapped, the cmma_d bits are cleared but the cmma_dirty_pages counter is never decremented:
arch/s390/kvm/dat.c:dat_reset_cmma_pte() {
...
pgste.cmma_d = 0;
...
}When migration is started again, __set_cmma_dirty_pte() unconditionally increments cmma_dirty_pages for every page that has its cmma_d bit flipped from 0 to 1, compounding the leaked counter:
arch/s390/kvm/gmap.c:__set_cmma_dirty_pte() {
...
if (!pgste.cmma_d) {
pgste.cmma_d = 1;
atomic64_inc(walk->priv);
}
...
}During migration, KVM reports this artificially inflated counter to userspace via args->remaining.
Once QEMU successfully retrieves all actual dirty pages, KVM will return count == 0 but remaining > 0 (due to the leak). Could this trap QEMU in an infinite migration polling loop waiting for pages that do not exist?
Quoted 4 lines
arch/s390/kvm/kvm-s390.c +0
-1- kvm->arch.migration_mode =1 ;kvm_s390_sync_request_broadcast (kvm,KVM_REQ_START_MIGRATION );return 0 ; }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=6
Claudio Imbrenda
maintainerPATCH v4 7/7 KVM: s390: selftests: Fix cmma selftest
In reply to Claudio ImbrendaThe existing cmma selftest depended on the host allocating page tables for all present memslots. Since the gmap rewrite, memory that is not accessed by the guest might not have page tables allocated yet.
This caused the test to fail due to a mismatch in the assertion.
Fix by having the guest access also the second half of the test memslot, thus guaranteeing that its page tables are present.
--- tools/testing/selftests/kvm/s390/cmma_test.c | 3 +++ 1 file changed, 3 insertions(+)
tools/testing/selftests/kvm/s390/cmma_test.c +3 -0
@@ -39,11 +39,14 @@ static void guest_do_one_essa(void) { asmvolatile (/* load TEST_DATA_START_GFN into r1 */ +" xgr 1,1\n" " llilf 1,%[start_gfn]\n" /* calculate the address from the gfn */ " sllg 1,1,12(0)\n" /* set the first page in TEST_DATA memslot to STABLE */ " .insn rrf,0xb9ab0000,2,1,1,0\n" +" agfi 1,0x100000\n" +" .insn rrf,0xb9ab0000,2,1,1,0\n" /* hypercall */ " diag 0,0,0x501\n" "0: j 0b"
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v4 7/7] KVM: s390: selftests: Fix cmma selftest
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] Outdated comments and function naming due to inline assembly changes. --
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Outdated comments and function naming due to inline assembly changes. --
commit 007cfe6c8e5c96330ecd4725d320fa59c950105a
Author: Claudio Imbrenda <[email protected]>
KVM: s390: selftests: Fix cmma selftest
The existing cmma selftest depended on the host allocating page tables for all present memslots. Since the gmap rewrite, memory that is not accessed by the guest might not have page tables allocated yet.
Quoted 6 lines
tools/testing/selftests/kvm/s390/cmma_test.c +0
-0@@ -39,11 +39,14 @@ static void guest_do_one_essa(void) {
tools/testing/selftests/kvm/s390/cmma_test.c static void guest_do_one_essa(void) · line 49
Quoted 10 lines
tools/testing/selftests/kvm/s390/cmma_test.c +3
-0asmvolatile (/* load TEST_DATA_START_GFN into r1 */ +" xgr 1,1\n" " llilf 1,%[start_gfn]\n" /* calculate the address from the gfn */ " sllg 1,1,12(0)\n" /* set the first page in TEST_DATA memslot to STABLE */ " .insn rrf,0xb9ab0000,2,1,1,0\n" +" agfi 1,0x100000\n" + " .insn rrf,0xb9ab0000,2,1,1,0\n"
[Severity: Low]
This isn't a bug, but with the addition of the second .insn instruction, the code now accesses a second page. Should the inline comment and the function name guest_do_one_essa (along with its docstring claiming it dirties exactly one page) be updated to reflect that multiple pages are now being touched?
Quoted 3 lines
tools/testing/selftests/kvm/s390/cmma_test.c +0
-0/* hypercall */ " diag 0,0,0x501\n" "0: j 0b"
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=7