[linus:master] [maple_tree] 280b792cac: will-it-scale.per_process_ops 6.0% regression

[Liam R. Howlett]

On 26/05/14 03:18PM, Oliver Sang wrote:
> hi, Liam,
> 
> On Wed, May 13, 2026 at 08:16:42PM -0400, Liam R. Howlett wrote:
> > On 26/05/13 03:40PM, kernel test robot wrote:
> > > 
> > > 
> > > Hello,
> > > 
> > > kernel test robot noticed a 6.0% regression of will-it-scale.per_process_ops on:
> > > 
> > > 
> > > commit: 280b792cac62ddadca2935766ca870b438c86323 ("maple_tree: use maple copy node for mas_wr_split()")
> > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
> > > 
> > > [still regression on linus/master      5d6919055dec134de3c40167a490f33c74c12581]
> > > [still regression on linux-next/master e98d21c170b01ddef366f023bbfcf6b31509fa83]
> > > 
> > > testcase: will-it-scale
> > > config: x86_64-rhel-9.4
> > > compiler: gcc-14
> > > test machine: 48 threads 2 sockets Intel(R) Xeon(R) CPU E5-2697 v2 @ 2.70GHz (Ivy Bridge-EP) with 64G memory
> > > parameters:
> > > 
> > > 	nr_task: 100%
> > > 	mode: process
> > > 	test: mmap2
> > > 	cpufreq_governor: performance
> > > 
> > > 
> > 
> > Thank you for the report.
> > 
> > 48 threads on a 2 socket E5-2697 v2 looks to be 12 cores (24 threads)
> > per cpu (so x2), or exactly one mmap2 process per hyperthread.
> 
> this is the cpu information:
> 
> Architecture:        x86_64
> CPU op-mode(s):      32-bit, 64-bit
> Byte Order:          Little Endian
> Address sizes:       46 bits physical, 48 bits virtual
> CPU(s):              48
> On-line CPU(s) list: 0-47
> Thread(s) per core:  2
> Core(s) per socket:  12
> Socket(s):           2
> NUMA node(s):        2
> Vendor ID:           GenuineIntel
> CPU family:          6
> Model:               62
> Model name:          Intel(R) Xeon(R) CPU E5-2697 v2 @ 2.70GHz
> Stepping:            4
> 
> > 
> > Is this across all process counts and peaks at 48, or just 48?
> 
> just 48. the run script is in
> https://download.01.org/0day-ci/archive/20260513/202605131554.92e7df6b-lkp@intel.com/repro-script
> 
> cd /lkp/benchmarks/will-it-scale
> python3 ./runtest.py mmap2 295 process 0 0 48
> 
> > Is this across many runs?
> 
> we run 6 times for both parent and this commit, the data looks stable
> 
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json:  "will-it-scale.per_process_ops": [
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    143595,
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    143474,
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    144104,
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    142796,
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    143081,
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-    143623
> 11e7f22f5e85058b09ca90e74002a3b82f50e940/matrix.json-  ],
> 
> 
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json:  "will-it-scale.per_process_ops": [
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    134451,
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    135089,
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    135080,
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    135039,
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    134082,
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-    135301
> 280b792cac62ddadca2935766ca870b438c86323/matrix.json-  ],
> 
> 
> > 
> > My testing didn't produce anything like this.  I'll have a look into
> > this when I can, but there isn't anything obvious that sticks out as a
> > likely cause.
> 
> if you want us to test any debug patch, it will be our great pleasure. thanks!

It looks like the result of the shape of the tree changing.  Can you try
the attached patch against Linus' tree?

Let me know if it'd be better to provide a copy based off 280b792cac62d
instead.

Thanks,
Liam
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-maple_tree-Restore-old-tree-layout-using-new-scatter.patch
Type: text/x-diff
Size: 22932 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/maple-tree/attachments/20260521/31371a38/attachment.bin>