[RFC PATCH 2/6] fs/resctrl: Do not update bandwidth from a failed counter read

[Jinjie Ruan <ruanjinjie01@gmail.com>]

From: Jinjie Ruan <ruanjinjie@huawei.com>

mbm_update_one_event() reads a monitoring group's counter once per
second and updates the bandwidth used by the MBA software controller
through mbm_bw_count().

When the read fails, rmid_read::err is set and the accumulated counter
value stays zero, but mbm_bw_count() runs anyway. Subtracting the
previous sample from zero wraps to a huge value, which becomes the
group's bandwidth; the controller then moves the control the wrong
way. The failure is reachable with assignable counters, where a group
has no counter for the event once the shared pool is exhausted.

Update the bandwidth only when the read succeeded. A failed read keeps
the previous bandwidth; the counter keeps accumulating, so the next
successful sample spans the failed interval and is larger. The
controller steps down once and recovers on the next round.

Fixes: 2c272fadb58b ("x86/resctrl: Compute memory bandwidth for all supported events")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Cc: <stable@vger.kernel.org> # v6.14+
---
 fs/resctrl/monitor.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/resctrl/monitor.c b/fs/resctrl/monitor.c
index dbc6770508e6..2319e8e619bd 100644
--- a/fs/resctrl/monitor.c
+++ b/fs/resctrl/monitor.c
@@ -788,10 +788,11 @@ static void mbm_update_one_event(struct rdt_resource *r, struct rdt_l3_mon_domai
 	__mon_event_count(rdtgrp, &rr);
 
 	/*
-	 * If the software controller is enabled, compute the
-	 * bandwidth for this event id.
+	 * If the software controller is enabled, compute the bandwidth for
+	 * this event id. Do not update it when the counter could not be read:
+	 * that would store a bogus zero sample as the previous bandwidth.
 	 */
-	if (is_mba_sc(NULL))
+	if (is_mba_sc(NULL) && !rr.err)
 		mbm_bw_count(rdtgrp, &rr);
 
 	if (rr.arch_mon_ctx)
-- 
2.43.0