PM: hibernate: make compression threads configurable

The number of compression/decompression threads has a direct impact on
hibernate image generation and resume latency. Using more threads can
reduce overall resume time, but on systems with fewer CPU cores it may
also introduce contention and reduce efficiency.

Performance was evaluated on an 8-core ARM system, averaged over 10 runs:

    Threads  Hibernate(s)  Resume(s)
    --------------------------------
       3         12.14       18.86
       4         12.28       17.48
       5         11.09       16.77
       6         11.08       16.44

With 5–6 threads, resume latency improves by approximately 12% compared
to the default 3-thread configuration, with negligible impact on
hibernate time.

Introduce a new kernel parameter `hibernate_compression_threads=` that
allows users and integrators to tune the number of
compression/decompression threads at boot. This provides a way to
balance performance and CPU utilization across a wide range of hardware
without recompiling the kernel.

Signed-off-by: Xueqin Luo <luoxueqin@kylinos.cn>
Link: https://patch.msgid.link/f24b3ca6416e230a515a154ed4c121d72a7e05a6.1761046167.git.luoxueqin@kylinos.cn
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

2 files changed, 31 insertions, 4 deletions

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 6c42061ca20e..46db3cbb838f 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -1907,6 +1907,16 @@
 			/sys/power/pm_test). Only available when CONFIG_PM_DEBUG
 			is set. Default value is 5.
 
+	hibernate_compression_threads=
+			[HIBERNATION]
+			Set the number of threads used for compressing or decompressing
+			hibernation images.
+
+			Format: <integer>
+			Default: 3
+			Minimum: 1
+			Example: hibernate_compression_threads=4
+
 	highmem=nn[KMG]	[KNL,BOOT,EARLY] forces the highmem zone to have an exact
 			size of <nn>. This works even on boxes that have no
 			highmem otherwise. This also works to reduce highmem
diff --git a/kernel/power/swap.c b/kernel/power/swap.c
index f8c13f5672ec..aa11576e92a9 100644
--- a/kernel/power/swap.c
+++ b/kernel/power/swap.c
@@ -519,8 +519,9 @@ static int swap_writer_finish(struct swap_map_handle *handle,
 				CMP_HEADER, PAGE_SIZE)
 #define CMP_SIZE	(CMP_PAGES * PAGE_SIZE)
 
-/* Maximum number of threads for compression/decompression. */
-#define CMP_THREADS	3
+/* Default number of threads for compression/decompression. */
+#define CMP_THREADS    3
+static unsigned int hibernate_compression_threads = CMP_THREADS;
 
 /* Minimum/maximum number of pages for read buffering. */
 #define CMP_MIN_RD_PAGES	1024
@@ -741,7 +742,7 @@ static int save_compressed_image(struct swap_map_handle *handle,
 	 * footprint.
 	 */
 	nr_threads = num_online_cpus() - 1;
-	nr_threads = clamp_val(nr_threads, 1, CMP_THREADS);
+	nr_threads = clamp_val(nr_threads, 1, hibernate_compression_threads);
 
 	page = (void *)__get_free_page(GFP_NOIO | __GFP_HIGH);
 	if (!page) {
@@ -1257,7 +1258,7 @@ static int load_compressed_image(struct swap_map_handle *handle,
 	 * footprint.
 	 */
 	nr_threads = num_online_cpus() - 1;
-	nr_threads = clamp_val(nr_threads, 1, CMP_THREADS);
+	nr_threads = clamp_val(nr_threads, 1, hibernate_compression_threads);
 
 	page = vmalloc_array(CMP_MAX_RD_PAGES, sizeof(*page));
 	if (!page) {
@@ -1697,3 +1698,19 @@ static int __init swsusp_header_init(void)
 }
 
 core_initcall(swsusp_header_init);
+
+static int __init hibernate_compression_threads_setup(char *str)
+{
+	int rc = kstrtouint(str, 0, &hibernate_compression_threads);
+
+	if (rc)
+		return rc;
+
+	if (hibernate_compression_threads < 1)
+		hibernate_compression_threads = CMP_THREADS;
+
+	return 1;
+
+}
+
+__setup("hibernate_compression_threads=", hibernate_compression_threads_setup);