From: Saurabh Sengar ssengar@linux.microsoft.com
On a x86 system under test with 1780 CPUs, topology_span_sane() takes around 8 seconds cumulatively for all the iterations. It is an expensive operation which does the sanity of non-NUMA topology masks.
CPU topology is not something which changes very frequently hence make this check optional for the systems where the topology is trusted and need faster bootup.
Restrict this to sched_verbose kernel cmdline option so that this penalty can be avoided for the systems who want to avoid it.
Cc: stable@vger.kernel.org Fixes: ccf74128d66c ("sched/topology: Assert non-NUMA topology masks don't (partially) overlap") Signed-off-by: Saurabh Sengar ssengar@linux.microsoft.com Co-developed-by: Naman Jain namjain@linux.microsoft.com Signed-off-by: Naman Jain namjain@linux.microsoft.com Tested-by: K Prateek Nayak kprateek.nayak@amd.com --- Changes since v3: https://lore.kernel.org/all/20250203114738.3109-1-namjain@linux.microsoft.co... - Minor typo correction in comment - Added Tested-by tag from Prateek for x86 Changes since v2: https://lore.kernel.org/all/1731922777-7121-1-git-send-email-ssengar@linux.m... - Use sched_debug() instead of using sched_debug_verbose variable directly (addressing Prateek's comment)
Changes since v1: https://lore.kernel.org/all/1729619853-2597-1-git-send-email-ssengar@linux.m... - Use kernel cmdline param instead of compile time flag.
Adding a link to the other patch which is under review. https://lore.kernel.org/lkml/20241031200431.182443-1-steve.wahl@hpe.com/ Above patch tries to optimize the topology sanity check, whereas this patch makes it optional. We believe both patches can coexist, as even with optimization, there will still be some performance overhead for this check.
--- kernel/sched/topology.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index c49aea8c1025..666f0a18cc6c 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2359,6 +2359,13 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl, { int i = cpu + 1;
+ /* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ + if (!sched_debug()) { + pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + __func__); + return true; + } + /* NUMA levels are allowed to overlap */ if (tl->flags & SDTL_OVERLAP) return true;
On 06/03/25 11:23, Naman Jain wrote:
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index c49aea8c1025..666f0a18cc6c 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2359,6 +2359,13 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl, { int i = cpu + 1;
- /* Skip the topology sanity check for non-debug, as it is a time-consuming operation */
- if (!sched_debug()) {
pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n",__func__);
FWIW I'm not against this change, however if you want to add messaging about sched_verbose I'd put that in e.g. sched_domain_debug() (as a print once like you've done here) with something along the lines of:
"Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it"
return true;- }
/* NUMA levels are allowed to overlap */ if (tl->flags & SDTL_OVERLAP) return true;-- 2.34.1
On 3/6/2025 10:18 PM, Valentin Schneider wrote:
On 06/03/25 11:23, Naman Jain wrote:
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index c49aea8c1025..666f0a18cc6c 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2359,6 +2359,13 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl, { int i = cpu + 1;
- /* Skip the topology sanity check for non-debug, as it is a time-consuming operation */
- if (!sched_debug()) {
pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n",__func__);FWIW I'm not against this change, however if you want to add messaging about sched_verbose I'd put that in e.g. sched_domain_debug() (as a print once like you've done here) with something along the lines of:
"Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it"
Thank you so much for reviewing. Please correct me if I misunderstood. Are you proposing below change?
--- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2361,7 +2361,7 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl,
/* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ if (!sched_debug()) { - pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + pr_info_once("%s: Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it\n", __func__); return true; }
Regards, Naman
return true;- }
/* NUMA levels are allowed to overlap */ if (tl->flags & SDTL_OVERLAP) return true;-- 2.34.1
Hello Naman,
On 3/7/2025 8:35 PM, Naman Jain wrote:
On 3/6/2025 10:18 PM, Valentin Schneider wrote:
On 06/03/25 11:23, Naman Jain wrote:
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index c49aea8c1025..666f0a18cc6c 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2359,6 +2359,13 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl, { int i = cpu + 1;
+ /* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ + if (!sched_debug()) { + pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + __func__);
FWIW I'm not against this change, however if you want to add messaging about sched_verbose I'd put that in e.g. sched_domain_debug() (as a print once like you've done here) with something along the lines of:
"Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it"
Thank you so much for reviewing. Please correct me if I misunderstood. Are you proposing below change?
--- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2361,7 +2361,7 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl,
/* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ if (!sched_debug()) { - pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + pr_info_once("%s: Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it\n", __func__); return true; }
I think Valentin meant moving the same pr_info_once() to the early exit case in sched_domain_debug() for "!sched_debug_verbose" to notify the user that sched_debug() is disabled and they can turn it on using "sched_verbose" as opposed to announcing it from topology_span_sane().
On 3/7/2025 9:11 PM, K Prateek Nayak wrote:
Hello Naman,
On 3/7/2025 8:35 PM, Naman Jain wrote:
On 3/6/2025 10:18 PM, Valentin Schneider wrote:
On 06/03/25 11:23, Naman Jain wrote:
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index c49aea8c1025..666f0a18cc6c 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2359,6 +2359,13 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl, { int i = cpu + 1;
+ /* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ + if (!sched_debug()) { + pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + __func__);
FWIW I'm not against this change, however if you want to add messaging about sched_verbose I'd put that in e.g. sched_domain_debug() (as a print once like you've done here) with something along the lines of:
"Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it"
Thank you so much for reviewing. Please correct me if I misunderstood. Are you proposing below change?
--- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2361,7 +2361,7 @@ static bool topology_span_sane(struct sched_domain_topology_level *tl,
/* Skip the topology sanity check for non-debug, as it is a time-consuming operation */ if (!sched_debug()) { - pr_info_once("%s: Skipping topology span sanity check. Use `sched_verbose` boot parameter to enable it.\n", + pr_info_once("%s: Scheduler topology debugging disabled, add 'sched_verbose' to the cmdline to enable it\n", __func__); return true; }
I think Valentin meant moving the same pr_info_once() to the early exit case in sched_domain_debug() for "!sched_debug_verbose" to notify the user that sched_debug() is disabled and they can turn it on using "sched_verbose" as opposed to announcing it from topology_span_sane().
Sure, thanks. Sent v5 with this change.
Regards, Naman
linux-stable-mirror@lists.linaro.org
-
K Prateek Nayak -
Naman Jain -
Valentin Schneider