rasdaemon icon indicating copy to clipboard operation
rasdaemon copied to clipboard

rasdaemon: introduce poison page statistics

Open winterddd opened this issue 9 months ago • 2 comments

An excessive number of poison pages can lead to memory fragmentation, which may degrade system performance. This patch introduces a threshold monitoring mechanism for poison pages. When the number of poison pages exceeds the predefined threshold, a warning is issued to alert administrators.

winterddd avatar Mar 26 '25 06:03 winterddd

More modify: add ras_poison_page_stat when occuring page offline

diff --git a/ras-page-isolation.c b/ras-page-isolation.c
index 2166f5c..f3b1696 100644
--- a/ras-page-isolation.c
+++ b/ras-page-isolation.c
@@ -349,6 +349,10 @@ static void page_offline(struct page_record *pr)
 
        log(TERM, LOG_INFO, "Result of offlining page at %#llx: %s\n",
            addr, page_state[pr->offlined]);
+
+#ifdef HAVE_POISON_PAGE_STAT
+       ras_poison_page_stat();
+#endif
 }

winterddd avatar Mar 31 '25 02:03 winterddd

LGTM, now. Thanks.

Reviewed-by: Shuai Xue [email protected]

axiqia avatar Apr 28 '25 08:04 axiqia

Merged, thanks!

mchehab avatar Nov 14 '25 12:11 mchehab