From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Dave Martin <Dave.Martin@arm.com>, Will Deacon <will@kernel.org>,
Sasha Levin <sashal@kernel.org>,
linux-arm-kernel@lists.infradead.org
Subject: [PATCH AUTOSEL 5.4 06/24] arm64/sve: Eliminate data races on sve_default_vl
Date: Tue, 23 Jun 2020 13:35:41 -0400 [thread overview]
Message-ID: <20200623173559.1355728-6-sashal@kernel.org> (raw)
In-Reply-To: <20200623173559.1355728-1-sashal@kernel.org>
From: Dave Martin <Dave.Martin@arm.com>
[ Upstream commit 1e570f512cbdc5e9e401ba640d9827985c1bea1e ]
sve_default_vl can be modified via the /proc/sys/abi/sve_default_vl
sysctl concurrently with use, and modified concurrently by multiple
threads.
Adding a lock for this seems overkill, and I don't want to think any
more than necessary, so just define wrappers using READ_ONCE()/
WRITE_ONCE().
This will avoid the possibility of torn accesses and repeated loads
and stores.
There's no evidence yet that this is going wrong in practice: this
is just hygiene. For generic sysctl users, it would be better to
build this kind of thing into the sysctl common code somehow.
Reported-by: Will Deacon <will@kernel.org>
Signed-off-by: Dave Martin <Dave.Martin@arm.com>
Link: https://lore.kernel.org/r/1591808590-20210-3-git-send-email-Dave.Martin@arm.com
[will: move set_sve_default_vl() inside #ifdef to squash allnoconfig warning]
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
arch/arm64/kernel/fpsimd.c | 25 ++++++++++++++++++-------
1 file changed, 18 insertions(+), 7 deletions(-)
diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index 1765e5284994f..d8895251a2aac 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -12,6 +12,7 @@
#include <linux/bug.h>
#include <linux/cache.h>
#include <linux/compat.h>
+#include <linux/compiler.h>
#include <linux/cpu.h>
#include <linux/cpu_pm.h>
#include <linux/kernel.h>
@@ -119,10 +120,20 @@ struct fpsimd_last_state_struct {
static DEFINE_PER_CPU(struct fpsimd_last_state_struct, fpsimd_last_state);
/* Default VL for tasks that don't set it explicitly: */
-static int sve_default_vl = -1;
+static int __sve_default_vl = -1;
+
+static int get_sve_default_vl(void)
+{
+ return READ_ONCE(__sve_default_vl);
+}
#ifdef CONFIG_ARM64_SVE
+static void set_sve_default_vl(int val)
+{
+ WRITE_ONCE(__sve_default_vl, val);
+}
+
/* Maximum supported vector length across all CPUs (initially poisoned) */
int __ro_after_init sve_max_vl = SVE_VL_MIN;
int __ro_after_init sve_max_virtualisable_vl = SVE_VL_MIN;
@@ -345,7 +356,7 @@ static int sve_proc_do_default_vl(struct ctl_table *table, int write,
loff_t *ppos)
{
int ret;
- int vl = sve_default_vl;
+ int vl = get_sve_default_vl();
struct ctl_table tmp_table = {
.data = &vl,
.maxlen = sizeof(vl),
@@ -362,7 +373,7 @@ static int sve_proc_do_default_vl(struct ctl_table *table, int write,
if (!sve_vl_valid(vl))
return -EINVAL;
- sve_default_vl = find_supported_vector_length(vl);
+ set_sve_default_vl(find_supported_vector_length(vl));
return 0;
}
@@ -869,7 +880,7 @@ void __init sve_setup(void)
* For the default VL, pick the maximum supported value <= 64.
* VL == 64 is guaranteed not to grow the signal frame.
*/
- sve_default_vl = find_supported_vector_length(64);
+ set_sve_default_vl(find_supported_vector_length(64));
bitmap_andnot(tmp_map, sve_vq_partial_map, sve_vq_map,
SVE_VQ_MAX);
@@ -890,7 +901,7 @@ void __init sve_setup(void)
pr_info("SVE: maximum available vector length %u bytes per vector\n",
sve_max_vl);
pr_info("SVE: default vector length %u bytes per vector\n",
- sve_default_vl);
+ get_sve_default_vl());
/* KVM decides whether to support mismatched systems. Just warn here: */
if (sve_max_virtualisable_vl < sve_max_vl)
@@ -1030,13 +1041,13 @@ void fpsimd_flush_thread(void)
* vector length configured: no kernel task can become a user
* task without an exec and hence a call to this function.
* By the time the first call to this function is made, all
- * early hardware probing is complete, so sve_default_vl
+ * early hardware probing is complete, so __sve_default_vl
* should be valid.
* If a bug causes this to go wrong, we make some noise and
* try to fudge thread.sve_vl to a safe value here.
*/
vl = current->thread.sve_vl_onexec ?
- current->thread.sve_vl_onexec : sve_default_vl;
+ current->thread.sve_vl_onexec : get_sve_default_vl();
if (WARN_ON(!sve_vl_valid(vl)))
vl = SVE_VL_MIN;
--
2.25.1
next prev parent reply other threads:[~2020-06-23 17:40 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-23 17:35 [PATCH AUTOSEL 5.4 01/24] sata_rcar: handle pm_runtime_get_sync failure cases Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 02/24] ata/libata: Fix usage of page address by page_address in ata_scsi_mode_select_xlat function Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 03/24] drm/amd/display: Use kfree() to free rgb_user in calculate_user_regamma_ramp() Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 04/24] riscv/atomic: Fix sign extension for RV64I Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 05/24] hwrng: ks-sa - Fix runtime PM imbalance on error Sasha Levin
2020-06-23 17:35 ` Sasha Levin [this message]
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 07/24] ibmvnic: Harden device login requests Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 08/24] net: alx: fix race condition in alx_remove Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 09/24] test_objagg: Fix potential memory leak in error handling Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 10/24] rocker: fix incorrect error handling in dma_rings_init Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 11/24] mvpp2: ethtool rxtx stats fix Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 12/24] pinctrl: qcom: spmi-gpio: fix warning about irq chip reusage Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 13/24] pinctrl: tegra: Use noirq suspend/resume callbacks Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 14/24] s390/ptrace: pass invalid syscall numbers to tracing Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 15/24] s390/ptrace: fix setting syscall number Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 16/24] s390/vdso: Use $(LD) instead of $(CC) to link vDSO Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 17/24] s390/vdso: fix vDSO clock_getres() Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 18/24] arm64: sve: Fix build failure when ARM64_SVE=y and SYSCTL=n Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 19/24] kbuild: improve cc-option to clean up all temporary files Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 20/24] recordmcount: support >64k sections Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 21/24] kprobes: Suppress the suspicious RCU warning on kprobes Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 22/24] blktrace: break out of blktrace setup on concurrent calls Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 23/24] block: update hctx map when use multiple maps Sasha Levin
2020-06-23 17:35 ` [PATCH AUTOSEL 5.4 24/24] RISC-V: Don't allow write+exec only page mapping request in mmap Sasha Levin
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=20200623173559.1355728-6-sashal@kernel.org \
--to=sashal@kernel.org \
--cc=Dave.Martin@arm.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=will@kernel.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