linux-kernel - [PATCH v7 27/29] KVM: s390: Enable 1M pages for gmap

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20260204150259.60425-28-imbrenda@linux.ibm.com>
Date: Wed,  4 Feb 2026 16:02:56 +0100
From: Claudio Imbrenda <imbrenda@...ux.ibm.com>
To: kvm@...r.kernel.org
Cc: linux-kernel@...r.kernel.org, linux-s390@...r.kernel.org,
        borntraeger@...ibm.com, frankja@...ux.ibm.com, nsg@...ux.ibm.com,
        nrb@...ux.ibm.com, seiden@...ux.ibm.com, gra@...ux.ibm.com,
        schlameuss@...ux.ibm.com, hca@...ux.ibm.com, svens@...ux.ibm.com,
        agordeev@...ux.ibm.com, gor@...ux.ibm.com, david@...nel.org,
        gerald.schaefer@...ux.ibm.com
Subject: [PATCH v7 27/29] KVM: s390: Enable 1M pages for gmap

While userspace is allowed to have pages of any size, the new gmap
would always use 4k pages to back the guest.

Enable 1M pages for gmap.

This allows 1M pages to be used to back a guest when userspace is using
1M pages for the corresponding addresses (e.g. THP or hugetlbfs).

Remove the limitation that disallowed having nested guests and
hugepages at the same time.

Signed-off-by: Claudio Imbrenda <imbrenda@...ux.ibm.com>
---
 arch/s390/kvm/gmap.c     | 2 +-
 arch/s390/kvm/kvm-s390.c | 6 +-----
 arch/s390/kvm/pv.c       | 3 +++
 3 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/arch/s390/kvm/gmap.c b/arch/s390/kvm/gmap.c
index fea1c66fcabe..da222962ef6d 100644
--- a/arch/s390/kvm/gmap.c
+++ b/arch/s390/kvm/gmap.c
@@ -620,7 +620,7 @@ static inline bool gmap_2g_allowed(struct gmap *gmap, gfn_t gfn)
 
 static inline bool gmap_1m_allowed(struct gmap *gmap, gfn_t gfn)
 {
-	return false;
+	return test_bit(GMAP_FLAG_ALLOW_HPAGE_1M, &gmap->flags);
 }
 
 int gmap_link(struct kvm_s390_mmu_cache *mc, struct gmap *gmap, struct guest_fault *f)
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index bde55761bf8a..ac7b5f56f0b5 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -851,6 +851,7 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
 			r = -EINVAL;
 		else {
 			r = 0;
+			set_bit(GMAP_FLAG_ALLOW_HPAGE_1M, &kvm->arch.gmap->flags);
 			/*
 			 * We might have to create fake 4k page
 			 * tables. To avoid that the hardware works on
@@ -5729,11 +5730,6 @@ static int __init kvm_s390_init(void)
 		return -ENODEV;
 	}
 
-	if (nested && hpage) {
-		pr_info("A KVM host that supports nesting cannot back its KVM guests with huge pages\n");
-		return -EINVAL;
-	}
-
 	for (i = 0; i < 16; i++)
 		kvm_s390_fac_base[i] |=
 			stfle_fac_list[i] & nonhyp_mask(i);
diff --git a/arch/s390/kvm/pv.c b/arch/s390/kvm/pv.c
index a48a8afd40df..461b413c76a3 100644
--- a/arch/s390/kvm/pv.c
+++ b/arch/s390/kvm/pv.c
@@ -721,6 +721,9 @@ int kvm_s390_pv_init_vm(struct kvm *kvm, u16 *rc, u16 *rrc)
 	uvcb.flags.ap_allow_instr = kvm->arch.model.uv_feat_guest.ap;
 	uvcb.flags.ap_instr_intr = kvm->arch.model.uv_feat_guest.ap_intr;
 
+	clear_bit(GMAP_FLAG_ALLOW_HPAGE_1M, &kvm->arch.gmap->flags);
+	gmap_split_huge_pages(kvm->arch.gmap);
+
 	cc = uv_call_sched(0, (u64)&uvcb);
 	*rc = uvcb.header.rc;
 	*rrc = uvcb.header.rrc;
-- 
2.52.0