The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 7559950aef1ab8792c50797c6c5c7c5150a02460 Mon Sep 17 00:00:00 2001
From: Suzuki K Poulose <suzuki.poulose(a)arm.com>
Date: Mon, 13 Jan 2020 23:30:20 +0000
Subject: [PATCH] arm64: cpufeature: Set the FP/SIMD compat HWCAP bits properly
We set the compat_elf_hwcap bits unconditionally on arm64 to
include the VFP and NEON support. However, the FP/SIMD unit
is optional on Arm v8 and thus could be missing. We already
handle this properly in the kernel, but still advertise to
the COMPAT applications that the VFP is available. Fix this
to make sure we only advertise when we really have them.
Fixes: 82e0191a1aa11abf ("arm64: Support systems without FP/ASIMD")
Cc: Will Deacon <will(a)kernel.org>
Cc: Mark Rutland <mark.rutland(a)arm.com>
Reviewed-by: Ard Biesheuvel <ardb(a)kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas(a)arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose(a)arm.com>
Signed-off-by: Will Deacon <will(a)kernel.org>
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 2fc9f18e2d2d..c008164b0848 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -32,9 +32,7 @@ static unsigned long elf_hwcap __read_mostly;
#define COMPAT_ELF_HWCAP_DEFAULT \
(COMPAT_HWCAP_HALF|COMPAT_HWCAP_THUMB|\
COMPAT_HWCAP_FAST_MULT|COMPAT_HWCAP_EDSP|\
- COMPAT_HWCAP_TLS|COMPAT_HWCAP_VFP|\
- COMPAT_HWCAP_VFPv3|COMPAT_HWCAP_VFPv4|\
- COMPAT_HWCAP_NEON|COMPAT_HWCAP_IDIV|\
+ COMPAT_HWCAP_TLS|COMPAT_HWCAP_IDIV|\
COMPAT_HWCAP_LPAE)
unsigned int compat_elf_hwcap __read_mostly = COMPAT_ELF_HWCAP_DEFAULT;
unsigned int compat_elf_hwcap2 __read_mostly;
@@ -1597,6 +1595,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
.match_list = list, \
}
+#define HWCAP_CAP_MATCH(match, cap_type, cap) \
+ { \
+ __HWCAP_CAP(#cap, cap_type, cap) \
+ .matches = match, \
+ }
+
#ifdef CONFIG_ARM64_PTR_AUTH
static const struct arm64_cpu_capabilities ptr_auth_hwcap_addr_matches[] = {
{
@@ -1670,8 +1674,35 @@ static const struct arm64_cpu_capabilities arm64_elf_hwcaps[] = {
{},
};
+#ifdef CONFIG_COMPAT
+static bool compat_has_neon(const struct arm64_cpu_capabilities *cap, int scope)
+{
+ /*
+ * Check that all of MVFR1_EL1.{SIMDSP, SIMDInt, SIMDLS} are available,
+ * in line with that of arm32 as in vfp_init(). We make sure that the
+ * check is future proof, by making sure value is non-zero.
+ */
+ u32 mvfr1;
+
+ WARN_ON(scope == SCOPE_LOCAL_CPU && preemptible());
+ if (scope == SCOPE_SYSTEM)
+ mvfr1 = read_sanitised_ftr_reg(SYS_MVFR1_EL1);
+ else
+ mvfr1 = read_sysreg_s(SYS_MVFR1_EL1);
+
+ return cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDSP_SHIFT) &&
+ cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDINT_SHIFT) &&
+ cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDLS_SHIFT);
+}
+#endif
+
static const struct arm64_cpu_capabilities compat_elf_hwcaps[] = {
#ifdef CONFIG_COMPAT
+ HWCAP_CAP_MATCH(compat_has_neon, CAP_COMPAT_HWCAP, COMPAT_HWCAP_NEON),
+ HWCAP_CAP(SYS_MVFR1_EL1, MVFR1_SIMDFMAC_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFPv4),
+ /* Arm v8 mandates MVFR0.FPDP == {0, 2}. So, piggy back on this for the presence of VFP support */
+ HWCAP_CAP(SYS_MVFR0_EL1, MVFR0_FPDP_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFP),
+ HWCAP_CAP(SYS_MVFR0_EL1, MVFR0_FPDP_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFPv3),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_PMULL),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_AES),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_SHA1_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_SHA1),
The patch below does not apply to the 4.19-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 7559950aef1ab8792c50797c6c5c7c5150a02460 Mon Sep 17 00:00:00 2001
From: Suzuki K Poulose <suzuki.poulose(a)arm.com>
Date: Mon, 13 Jan 2020 23:30:20 +0000
Subject: [PATCH] arm64: cpufeature: Set the FP/SIMD compat HWCAP bits properly
We set the compat_elf_hwcap bits unconditionally on arm64 to
include the VFP and NEON support. However, the FP/SIMD unit
is optional on Arm v8 and thus could be missing. We already
handle this properly in the kernel, but still advertise to
the COMPAT applications that the VFP is available. Fix this
to make sure we only advertise when we really have them.
Fixes: 82e0191a1aa11abf ("arm64: Support systems without FP/ASIMD")
Cc: Will Deacon <will(a)kernel.org>
Cc: Mark Rutland <mark.rutland(a)arm.com>
Reviewed-by: Ard Biesheuvel <ardb(a)kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas(a)arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose(a)arm.com>
Signed-off-by: Will Deacon <will(a)kernel.org>
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 2fc9f18e2d2d..c008164b0848 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -32,9 +32,7 @@ static unsigned long elf_hwcap __read_mostly;
#define COMPAT_ELF_HWCAP_DEFAULT \
(COMPAT_HWCAP_HALF|COMPAT_HWCAP_THUMB|\
COMPAT_HWCAP_FAST_MULT|COMPAT_HWCAP_EDSP|\
- COMPAT_HWCAP_TLS|COMPAT_HWCAP_VFP|\
- COMPAT_HWCAP_VFPv3|COMPAT_HWCAP_VFPv4|\
- COMPAT_HWCAP_NEON|COMPAT_HWCAP_IDIV|\
+ COMPAT_HWCAP_TLS|COMPAT_HWCAP_IDIV|\
COMPAT_HWCAP_LPAE)
unsigned int compat_elf_hwcap __read_mostly = COMPAT_ELF_HWCAP_DEFAULT;
unsigned int compat_elf_hwcap2 __read_mostly;
@@ -1597,6 +1595,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
.match_list = list, \
}
+#define HWCAP_CAP_MATCH(match, cap_type, cap) \
+ { \
+ __HWCAP_CAP(#cap, cap_type, cap) \
+ .matches = match, \
+ }
+
#ifdef CONFIG_ARM64_PTR_AUTH
static const struct arm64_cpu_capabilities ptr_auth_hwcap_addr_matches[] = {
{
@@ -1670,8 +1674,35 @@ static const struct arm64_cpu_capabilities arm64_elf_hwcaps[] = {
{},
};
+#ifdef CONFIG_COMPAT
+static bool compat_has_neon(const struct arm64_cpu_capabilities *cap, int scope)
+{
+ /*
+ * Check that all of MVFR1_EL1.{SIMDSP, SIMDInt, SIMDLS} are available,
+ * in line with that of arm32 as in vfp_init(). We make sure that the
+ * check is future proof, by making sure value is non-zero.
+ */
+ u32 mvfr1;
+
+ WARN_ON(scope == SCOPE_LOCAL_CPU && preemptible());
+ if (scope == SCOPE_SYSTEM)
+ mvfr1 = read_sanitised_ftr_reg(SYS_MVFR1_EL1);
+ else
+ mvfr1 = read_sysreg_s(SYS_MVFR1_EL1);
+
+ return cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDSP_SHIFT) &&
+ cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDINT_SHIFT) &&
+ cpuid_feature_extract_unsigned_field(mvfr1, MVFR1_SIMDLS_SHIFT);
+}
+#endif
+
static const struct arm64_cpu_capabilities compat_elf_hwcaps[] = {
#ifdef CONFIG_COMPAT
+ HWCAP_CAP_MATCH(compat_has_neon, CAP_COMPAT_HWCAP, COMPAT_HWCAP_NEON),
+ HWCAP_CAP(SYS_MVFR1_EL1, MVFR1_SIMDFMAC_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFPv4),
+ /* Arm v8 mandates MVFR0.FPDP == {0, 2}. So, piggy back on this for the presence of VFP support */
+ HWCAP_CAP(SYS_MVFR0_EL1, MVFR0_FPDP_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFP),
+ HWCAP_CAP(SYS_MVFR0_EL1, MVFR0_FPDP_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP, COMPAT_HWCAP_VFPv3),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_PMULL),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_AES),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_SHA1_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_SHA1),
While certain modeset operations on gv100+ need us to temporarily
disable the LUT, we make the mistake of sometimes neglecting to
reprogram the LUT after such modesets. In particular, moving a head from
one encoder to another seems to trigger this quite often. GV100+ is very
picky about having a LUT in most scenarios, so this causes the display
engine to hang with the following error code:
disp: chid 1 stat 00005080 reason 5 [INVALID_STATE] mthd 0200 data
00000001 code 0000002d)
So, fix this by always re-programming the LUT if we're clearing it in a
state where the wndw is still visible, and has a XLUT handle programmed.
Signed-off-by: Lyude Paul <lyude(a)redhat.com>
Fixes: facaed62b4cb ("drm/nouveau/kms/gv100: initial support")
Cc: <stable(a)vger.kernel.org> # v4.18+
---
drivers/gpu/drm/nouveau/dispnv50/wndw.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/gpu/drm/nouveau/dispnv50/wndw.c b/drivers/gpu/drm/nouveau/dispnv50/wndw.c
index 890315291b01..bb737f9281e6 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/wndw.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/wndw.c
@@ -458,6 +458,8 @@ nv50_wndw_atomic_check(struct drm_plane *plane, struct drm_plane_state *state)
asyw->clr.ntfy = armw->ntfy.handle != 0;
asyw->clr.sema = armw->sema.handle != 0;
asyw->clr.xlut = armw->xlut.handle != 0;
+ if (asyw->clr.xlut && asyw->visible)
+ asyw->set.xlut = asyw->xlut.handle != 0;
asyw->clr.csc = armw->csc.valid;
if (wndw->func->image_clr)
asyw->clr.image = armw->image.handle[0] != 0;
--
2.24.1
This just limits the BPC for MST connectors to a maximum of 8 from
nv50_mstc_get_modes(), instead of doing so during
nv50_msto_atomic_check(). This doesn't introduce any functional changes
yet (other then userspace now lying about the max bpc, but we can't
support that yet anyway so meh). But, we'll need this in a moment so
that we can share mode validation between SST and MST which will fix
some real world issues.
Signed-off-by: Lyude Paul <lyude(a)redhat.com>
Cc: stable(a)vger.kernel.org
---
drivers/gpu/drm/nouveau/dispnv50/disp.c | 25 ++++++++++++++-----------
1 file changed, 14 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index 32a1c4221f1e..766b8e80a8f5 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -903,15 +903,9 @@ nv50_msto_atomic_check(struct drm_encoder *encoder,
if (!state->duplicated) {
const int clock = crtc_state->adjusted_mode.clock;
- /*
- * XXX: Since we don't use HDR in userspace quite yet, limit
- * the bpc to 8 to save bandwidth on the topology. In the
- * future, we'll want to properly fix this by dynamically
- * selecting the highest possible bpc that would fit in the
- * topology
- */
- asyh->or.bpc = min(connector->display_info.bpc, 8U);
- asyh->dp.pbn = drm_dp_calc_pbn_mode(clock, asyh->or.bpc * 3, false);
+ asyh->or.bpc = connector->display_info.bpc;
+ asyh->dp.pbn = drm_dp_calc_pbn_mode(clock, asyh->or.bpc * 3,
+ false);
}
slots = drm_dp_atomic_find_vcpi_slots(state, &mstm->mgr, mstc->port,
@@ -1071,8 +1065,17 @@ nv50_mstc_get_modes(struct drm_connector *connector)
if (mstc->edid)
ret = drm_add_edid_modes(&mstc->connector, mstc->edid);
- if (!mstc->connector.display_info.bpc)
- mstc->connector.display_info.bpc = 8;
+ /*
+ * XXX: Since we don't use HDR in userspace quite yet, limit the bpc
+ * to 8 to save bandwidth on the topology. In the future, we'll want
+ * to properly fix this by dynamically selecting the highest possible
+ * bpc that would fit in the topology
+ */
+ if (connector->display_info.bpc)
+ connector->display_info.bpc =
+ clamp(connector->display_info.bpc, 6U, 8U);
+ else
+ connector->display_info.bpc = 8;
if (mstc->native)
drm_mode_destroy(mstc->connector.dev, mstc->native);
--
2.24.1