From: Andrew Jones <drjones@redhat.com>
To: qemu-devel@nongnu.org, qemu-arm@nongnu.org
Cc: peter.maydell@linaro.org, richard.henderson@linaro.org,
armbru@redhat.com, abologna@redhat.com, alex.bennee@linaro.org,
Dave.Martin@arm.com
Subject: [Qemu-devel] [PATCH 12/13] target/arm/kvm: max cpu: Add support for sve-vls-map
Date: Sun, 12 May 2019 10:36:23 +0200 [thread overview]
Message-ID: <20190512083624.8916-13-drjones@redhat.com> (raw)
In-Reply-To: <20190512083624.8916-1-drjones@redhat.com>
The max cpu type can have its SVE vector lengths explicitly set
with the sve-vls-map property. This patch allows that property
to work when KVM is in use. The map must conform to additional
constraints for KVM which are checked at vcpu init.
Signed-off-by: Andrew Jones <drjones@redhat.com>
---
target/arm/cpu64.c | 7 +++---
target/arm/kvm64.c | 56 +++++++++++++++++++++++++++++++++++++++++++---
2 files changed, 56 insertions(+), 7 deletions(-)
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 9ac702d54136..94f3dd5b51e5 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -310,7 +310,7 @@ static void cpu_set_sve_vls_map(Object *obj, Visitor *v, const char *name,
error_setg(&err, "SVE vector length map has unsupported lengths");
error_append_hint(&err, "Valid vector lengths in range [1-%d]\n",
ARM_MAX_VQ);
- } else if (cpu->sve_max_vq != ARM_MAX_VQ &&
+ } else if (cpu->sve_max_vq != ARM_MAX_VQ && cpu->sve_max_vq != -1 &&
cpu->sve_max_vq != arm_cpu_fls64(cpu->sve_vls_map)) {
/*
* If the user provides both sve-max-vq and sve-vls-map, with
@@ -433,13 +433,12 @@ static void aarch64_max_initfn(Object *obj)
#endif
cpu->sve_max_vq = ARM_MAX_VQ;
-
- object_property_add(obj, "sve-vls-map", "uint64", cpu_get_sve_vls_map,
- cpu_set_sve_vls_map, NULL, NULL, &error_fatal);
}
object_property_add(obj, "sve-max-vq", "uint32", cpu_max_get_sve_vq,
cpu_max_set_sve_vq, NULL, NULL, &error_fatal);
+ object_property_add(obj, "sve-vls-map", "uint64", cpu_get_sve_vls_map,
+ cpu_set_sve_vls_map, NULL, NULL, &error_fatal);
}
struct ARMCPUInfo {
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
index 11c6334a7c08..5506f019c190 100644
--- a/target/arm/kvm64.c
+++ b/target/arm/kvm64.c
@@ -685,9 +685,9 @@ int kvm_arch_init_vcpu(CPUState *cs)
} else {
unset_feature(&env->features, ARM_FEATURE_PMU);
}
- if (cpu->sve_max_vq) {
+ if (cpu->sve_max_vq || cpu->sve_vls_map) {
if (!kvm_check_extension(cs->kvm_state, KVM_CAP_ARM_SVE)) {
- if (cpu->sve_max_vq == -1) {
+ if (cpu->sve_max_vq == -1 && !cpu->sve_vls_map) {
cpu->sve_max_vq = 0;
} else {
error_report("This KVM host does not support SVE");
@@ -704,12 +704,62 @@ int kvm_arch_init_vcpu(CPUState *cs)
return ret;
}
- if (cpu->sve_max_vq) {
+ if (cpu->sve_max_vq || cpu->sve_vls_map) {
uint64_t sve_vls[KVM_ARM64_SVE_VLS_WORDS];
ret = kvm_arm_get_sve_vls(cs, sve_vls);
if (ret < 0) {
return ret;
}
+ if (cpu->sve_vls_map) {
+ uint64_t ovls;
+ int i;
+
+ /*
+ * We currently only support a single VLS word, as that should
+ * be sufficient for some time (vq=64 means a 8192-bit vector
+ * and KVM currently only supports up to 2048-bit vectors).
+ * The choice to only support a single word for now is due to
+ * the need to input it on the command line. It's much simpler
+ * to input a word as a cpu property than an array of words.
+ * So for now just warn if we detect our assumption was wrong.
+ */
+ for (i = 1; i < KVM_ARM64_SVE_VLS_WORDS; ++i) {
+ if (sve_vls[i]) {
+ warn_report("KVM supports vector lengths larger than "
+ "sve-vls-map can select");
+ sve_vls[i] = 0;
+ }
+ }
+
+ ovls = sve_vls[0];
+ sve_vls[0] = cpu->sve_vls_map;
+
+ if (cpu->sve_vls_map & ~ovls) {
+ error_report("sve-vls-map=0x%lx is not valid on this host "
+ "which supports 0x%lx", cpu->sve_vls_map, ovls);
+ return -EINVAL;
+ }
+
+ i = arm_cpu_fls64(cpu->sve_vls_map);
+ if (cpu->sve_max_vq && cpu->sve_max_vq != -1 &&
+ cpu->sve_max_vq != i) {
+ error_report("sve-vls-map and sve-max-vq are inconsistent");
+ return -EINVAL;
+ }
+ cpu->sve_max_vq = i;
+
+ /*
+ * sve-vls-map must have all the same vector lengths up to its
+ * max vq that the host supports.
+ */
+ if (cpu->sve_vls_map != (ovls & (BIT_MASK(cpu->sve_max_vq) - 1))) {
+ error_report("sve-vls-map=0x%lx is not valid on this host "
+ "which supports 0x%lx", cpu->sve_vls_map, ovls);
+ error_printf("All host vector lengths up to %d must also "
+ "be selected.\n", cpu->sve_max_vq);
+ return -EINVAL;
+ }
+ }
if (cpu->sve_max_vq == -1) {
cpu->sve_max_vq = ret;
} else if (cpu->sve_max_vq > ret) {
--
2.20.1
next prev parent reply other threads:[~2019-05-12 8:43 UTC|newest]
Thread overview: 71+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-12 8:36 [Qemu-devel] [PATCH 00/13] target/arm/kvm: enable SVE in guests Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 01/13] target/arm/kvm64: fix error returns Andrew Jones
2019-06-05 7:20 ` Auger Eric
2019-05-12 8:36 ` [Qemu-devel] [PATCH 02/13] update-linux-headers: Add sve_context.h to asm-arm64 Andrew Jones
2019-06-05 7:21 ` Auger Eric
2019-06-05 7:30 ` Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 03/13] HACK: linux header update Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 04/13] target/arm/kvm: Move the get/put of fpsimd registers out Andrew Jones
2019-06-05 7:15 ` Auger Eric
2019-06-05 7:27 ` Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 05/13] target/arm/kvm: Add kvm_arch_get/put_sve Andrew Jones
2019-05-13 12:31 ` Dave Martin
2019-05-13 13:55 ` Andrew Jones
2019-05-13 15:31 ` Dave Martin
2019-05-13 15:40 ` Peter Maydell
2019-05-13 16:05 ` Dave Martin
2019-05-13 16:40 ` Richard Henderson
2019-05-13 18:14 ` Andrew Jones
2019-05-13 18:31 ` Richard Henderson
2019-05-13 12:43 ` Dave Martin
2019-05-13 14:07 ` Andrew Jones
2019-05-13 14:39 ` Dave Martin
2019-05-13 16:58 ` Richard Henderson
2019-05-14 9:10 ` Dave Martin
2019-05-12 8:36 ` [Qemu-devel] [PATCH 06/13] target/arm/kvm: max cpu: Enable SVE when available Andrew Jones
2019-06-05 9:09 ` Auger Eric
2019-06-05 11:04 ` Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 07/13] target/arm/kvm: max cpu: Allow sve max vector length setting Andrew Jones
2019-05-13 17:19 ` Richard Henderson
2019-05-13 18:19 ` Andrew Jones
2019-06-06 8:30 ` Auger Eric
2019-06-06 8:53 ` Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 08/13] target/arm/monitor: Add query-sve-vector-lengths Andrew Jones
2019-05-13 16:12 ` Markus Armbruster
2019-05-13 18:30 ` Andrew Jones
2019-05-14 5:32 ` Markus Armbruster
2019-05-12 8:36 ` [Qemu-devel] [PATCH 09/13] target/arm/kvm: Export kvm_arm_get_sve_vls Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 10/13] target/arm/monitor: kvm: only return valid sve vector sets Andrew Jones
2019-05-12 8:36 ` [Qemu-devel] [PATCH 11/13] target/arm/cpu64: max cpu: Introduce sve-vls-map Andrew Jones
2019-05-13 11:26 ` Dave Martin
2019-05-13 12:30 ` Andrew Jones
2019-05-13 12:41 ` Dave Martin
2019-05-13 12:57 ` Andrew Jones
2019-05-13 13:12 ` Dave Martin
2019-05-13 13:45 ` Andrew Jones
2019-05-13 14:35 ` Dave Martin
2019-05-13 15:25 ` Markus Armbruster
2019-05-13 18:31 ` Andrew Jones
2019-05-12 8:36 ` Andrew Jones [this message]
2019-05-12 8:36 ` [Qemu-devel] [PATCH 13/13] target/arm/kvm: host cpu: Add support for sve-vls-map Andrew Jones
2019-05-13 15:37 ` Markus Armbruster
2019-05-13 18:33 ` Andrew Jones
2019-05-13 9:32 ` [Qemu-devel] [PATCH 00/13] target/arm/kvm: enable SVE in guests Andrea Bolognani
2019-05-13 11:15 ` Dave Martin
2019-05-13 12:38 ` Andrew Jones
2019-05-13 12:50 ` Dave Martin
2019-05-13 12:36 ` Andrew Jones
2019-05-14 12:29 ` Andrea Bolognani
2019-05-14 12:53 ` Andrew Jones
2019-05-14 16:03 ` Andrea Bolognani
2019-05-14 20:14 ` Richard Henderson
2019-05-15 8:03 ` Andrea Bolognani
2019-05-15 11:14 ` Dave Martin
2019-05-15 11:28 ` Andrea Bolognani
2019-05-15 12:47 ` Dave Martin
2019-05-15 9:15 ` Andrew Jones
2019-05-13 9:52 ` Peter Maydell
2019-05-13 12:43 ` Andrew Jones
2019-05-13 18:46 ` Richard Henderson
2019-05-13 19:16 ` Andrew Jones
2019-05-14 9:05 ` Peter Maydell
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=20190512083624.8916-13-drjones@redhat.com \
--to=drjones@redhat.com \
--cc=Dave.Martin@arm.com \
--cc=abologna@redhat.com \
--cc=alex.bennee@linaro.org \
--cc=armbru@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.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).