[PATCH] test_xarray: fix soft lockup for advanced-api tests

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



The new adanced API tests want to vet the xarray API is doing what it
promises by manually iterating over a set of possible indexes on its
own, and using a query operation which holds the RCU lock and then
releases it. So it is not using the helper loop options which xarray
provides on purpose. Any loop which iterates over 1 million entries
(which is possible with order 20, so emulating say a 4 GiB block size)
to just to rcu lock and unlock will eventually end up triggering a soft
lockup on systems which don't preempt, and have lock provin and RCU
prooving enabled.

xarray users already use XA_CHECK_SCHED for loops which may take a long
time, in our case we don't want to RCU unlock and lock as the caller
does that already, but rather just force a schedule every XA_CHECK_SCHED
iterations since the test is trying to not trust and rather test that
xarray is doing the right thing.

[0] https://lkml.kernel.org/r/202402071613.70f28243-lkp@xxxxxxxxx

Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
Signed-off-by: Luis Chamberlain <mcgrof@xxxxxxxxxx>
---
 lib/test_xarray.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/lib/test_xarray.c b/lib/test_xarray.c
index d4e55b4867dc..ac162025cc59 100644
--- a/lib/test_xarray.c
+++ b/lib/test_xarray.c
@@ -781,6 +781,7 @@ static noinline void *test_get_entry(struct xarray *xa, unsigned long index)
 {
 	XA_STATE(xas, xa, index);
 	void *p;
+	static unsigned int i = 0;
 
 	rcu_read_lock();
 repeat:
@@ -790,6 +791,17 @@ static noinline void *test_get_entry(struct xarray *xa, unsigned long index)
 		goto repeat;
 	rcu_read_unlock();
 
+	/*
+	 * This is not part of the page cache, this selftest is pretty
+	 * aggressive and does not want to trust the xarray API but rather
+	 * test it, and for order 20 (4 GiB block size) we can loop over
+	 * over a million entries which can cause a soft lockup. Page cache
+	 * APIs won't be stupid, proper page cache APIs loop over the proper
+	 * order so when using a larger order we skip shared entries.
+	 */
+	if (++i % XA_CHECK_SCHED == 0)
+		schedule();
+
 	return p;
 }
 
-- 
2.42.0





[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux