From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Cc: lersek@redhat.com, kraxel@redhat.com, mst@redhat.com
Subject: [Qemu-devel] [PATCH v2 04/23] target-i386: mask NMIs on entry to SMM
Date: Wed, 3 Jun 2015 19:08:29 +0200 [thread overview]
Message-ID: <1433351328-23326-5-git-send-email-pbonzini@redhat.com> (raw)
In-Reply-To: <1433351328-23326-1-git-send-email-pbonzini@redhat.com>
QEMU is not blocking NMIs on entry to SMM. Implementing this has to
cover a few corner cases, because:
- NMIs can then be enabled by an IRET instruction and there
is no mechanism to _set_ the "NMIs masked" flag on exit from SMM:
"A special case can occur if an SMI handler nests inside an NMI handler
and then another NMI occurs. [...] When the processor enters SMM while
executing an NMI handler, the processor saves the SMRAM state save map
but does not save the attribute to keep NMI interrupts disabled.
- However, there is some hidden state, because "If NMIs were blocked
before the SMI occurred [and no IRET is executed while in SMM], they
are blocked after execution of RSM." This is represented by the new
HF2_SMM_INSIDE_NMI_MASK bit. If it is zero, NMIs are _unblocked_
on exit from RSM.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
target-i386/cpu.h | 20 +++++++++++---------
target-i386/smm_helper.c | 9 +++++++++
2 files changed, 20 insertions(+), 9 deletions(-)
diff --git a/target-i386/cpu.h b/target-i386/cpu.h
index 9f57fe9..4510ae7 100644
--- a/target-i386/cpu.h
+++ b/target-i386/cpu.h
@@ -180,15 +180,17 @@
/* hflags2 */
-#define HF2_GIF_SHIFT 0 /* if set CPU takes interrupts */
-#define HF2_HIF_SHIFT 1 /* value of IF_MASK when entering SVM */
-#define HF2_NMI_SHIFT 2 /* CPU serving NMI */
-#define HF2_VINTR_SHIFT 3 /* value of V_INTR_MASKING bit */
-
-#define HF2_GIF_MASK (1 << HF2_GIF_SHIFT)
-#define HF2_HIF_MASK (1 << HF2_HIF_SHIFT)
-#define HF2_NMI_MASK (1 << HF2_NMI_SHIFT)
-#define HF2_VINTR_MASK (1 << HF2_VINTR_SHIFT)
+#define HF2_GIF_SHIFT 0 /* if set CPU takes interrupts */
+#define HF2_HIF_SHIFT 1 /* value of IF_MASK when entering SVM */
+#define HF2_NMI_SHIFT 2 /* CPU serving NMI */
+#define HF2_VINTR_SHIFT 3 /* value of V_INTR_MASKING bit */
+#define HF2_SMM_INSIDE_NMI_SHIFT 4 /* CPU serving SMI nested inside NMI */
+
+#define HF2_GIF_MASK (1 << HF2_GIF_SHIFT)
+#define HF2_HIF_MASK (1 << HF2_HIF_SHIFT)
+#define HF2_NMI_MASK (1 << HF2_NMI_SHIFT)
+#define HF2_VINTR_MASK (1 << HF2_VINTR_SHIFT)
+#define HF2_SMM_INSIDE_NMI_MASK (1 << HF2_SMM_INSIDE_NMI_SHIFT)
#define CR0_PE_SHIFT 0
#define CR0_MP_SHIFT 1
diff --git a/target-i386/smm_helper.c b/target-i386/smm_helper.c
index b9971b6..6207c3a 100644
--- a/target-i386/smm_helper.c
+++ b/target-i386/smm_helper.c
@@ -52,6 +52,11 @@ void do_smm_enter(X86CPU *cpu)
log_cpu_state_mask(CPU_LOG_INT, CPU(cpu), CPU_DUMP_CCOP);
env->hflags |= HF_SMM_MASK;
+ if (env->hflags2 & HF2_NMI_MASK) {
+ env->hflags2 |= HF2_SMM_INSIDE_NMI_MASK;
+ } else {
+ env->hflags2 |= HF2_NMI_MASK;
+ }
cpu_smm_update(env);
sm_state = env->smbase + 0x8000;
@@ -307,6 +312,10 @@ void helper_rsm(CPUX86State *env)
env->smbase = x86_ldl_phys(cs, sm_state + 0x7ef8) & ~0x7fff;
}
#endif
+ if ((env->hflags2 & HF2_SMM_INSIDE_NMI_MASK) == 0) {
+ env->hflags2 &= ~HF2_NMI_MASK;
+ }
+ env->hflags2 &= ~HF2_SMM_INSIDE_NMI_MASK;
env->hflags &= ~HF_SMM_MASK;
cpu_smm_update(env);
--
2.4.1
next prev parent reply other threads:[~2015-06-03 17:09 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-03 17:08 [Qemu-devel] [PATCH v2 00/23] SMM improvements (chipset and TCG parts) Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 01/23] target-i386: introduce cpu_get_mem_attrs Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 02/23] target-i386: Use correct memory attributes for memory accesses Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 03/23] target-i386: Use correct memory attributes for ioport accesses Paolo Bonzini
2015-06-03 17:08 ` Paolo Bonzini [this message]
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 05/23] target-i386: set G=1 in SMM big real mode selectors Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 06/23] target-i386: wake up processors that receive an SMI Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 07/23] pflash_cfi01: change big-endian property to BIT type Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 08/23] pflash_cfi01: change to new-style MMIO accessors Paolo Bonzini
2015-06-04 6:19 ` Peter Crosthwaite
2015-06-04 8:02 ` Paolo Bonzini
2015-06-04 12:51 ` Laszlo Ersek
2015-06-09 18:08 ` Richard Henderson
2015-06-09 18:47 ` Michael S. Tsirkin
2015-06-17 7:56 ` Paolo Bonzini
2015-06-17 8:22 ` Markus Armbruster
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 09/23] pflash_cfi01: add secure property Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 10/23] vl: allow full-blown QemuOpts syntax for -global Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 11/23] qom: add object_property_add_const_link Paolo Bonzini
2015-06-04 6:33 ` Peter Crosthwaite
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 12/23] vl: run "late" notifiers immediately Paolo Bonzini
2015-06-04 6:39 ` Peter Crosthwaite
2015-06-04 8:03 ` Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 13/23] target-i386: create a separate AddressSpace for each CPU Paolo Bonzini
2015-06-03 17:58 ` Peter Crosthwaite
2015-06-04 8:02 ` Paolo Bonzini
2015-06-04 12:48 ` Laszlo Ersek
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 14/23] hw/i386: add a separate region that tracks the SMRAME bit Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 15/23] target-i386: use memory API to implement SMRAM Paolo Bonzini
2015-06-04 7:19 ` Peter Crosthwaite
2015-06-04 8:05 ` Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 16/23] hw/i386: remove smram_update Paolo Bonzini
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 17/23] q35: implement high SMRAM Paolo Bonzini
2015-06-04 12:50 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 18/23] q35: fix ESMRAMC default Paolo Bonzini
2015-06-04 12:51 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 19/23] q35: add config space wmask for SMRAM and ESMRAMC Paolo Bonzini
2015-06-04 12:51 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 20/23] q35: implement SMRAM.D_LCK Paolo Bonzini
2015-06-04 12:51 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 21/23] q35: add test for SMRAM.D_LCK Paolo Bonzini
2015-06-04 12:52 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 22/23] q35: implement TSEG Paolo Bonzini
2015-06-04 12:53 ` Michael S. Tsirkin
2015-06-03 17:08 ` [Qemu-devel] [PATCH v2 23/23] ich9: implement SMI_LOCK Paolo Bonzini
2015-06-04 12:53 ` Michael S. Tsirkin
2015-06-03 17:41 ` [Qemu-devel] [PATCH v2 00/23] SMM improvements (chipset and TCG parts) Laszlo Ersek
2015-06-03 17:44 ` Paolo Bonzini
2015-06-04 12:54 ` Michael S. Tsirkin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1433351328-23326-5-git-send-email-pbonzini@redhat.com \
--to=pbonzini@redhat.com \
--cc=kraxel@redhat.com \
--cc=lersek@redhat.com \
--cc=mst@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).