diff mbox series

tests: rrm: sleep between the many good updates

Message ID 1579086597-I3537c19232bc4e5f6f0a415deb332d97404d512e@changeid
State Superseded
Headers show
Series tests: rrm: sleep between the many good updates | expand

Commit Message

Johannes Berg Jan. 15, 2020, 11:09 a.m. UTC 
From: Johannes Berg <johannes.berg@intel.com>

These are all going to the kernel, and with powersave now
supported in hwsim, the frames are not transmitted immediately
but queued up. But the kernel has a limit of 64 queued up
frames when TX status ACK is requested, so this test fails.
Give it a minimal amount of time between frames so that they
can actually go out - 10ms for each means we get only 10 in
each beacon period before the other side wakes up and pulls
them, so we don't hit the 64 limit.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
---
 tests/hwsim/test_rrm.py | 1 +
 1 file changed, 1 insertion(+)

Comments

Johannes Berg Jan. 15, 2020, 11:27 a.m. UTC | #1 
On Wed, 2020-01-15 at 12:09 +0100, Johannes Berg wrote:
> From: Johannes Berg <johannes.berg@intel.com>
> 
> These are all going to the kernel, and with powersave now
> supported in hwsim, the frames are not transmitted immediately
> but queued up. But the kernel has a limit of 64 queued up
> frames when TX status ACK is requested, so this test fails.
> Give it a minimal amount of time between frames so that they
> can actually go out - 10ms for each means we get only 10 in
> each beacon period before the other side wakes up and pulls
> them, so we don't hit the 64 limit.

I'll fix the kernel too to increase that limit to 8192, since if this
happens on an AP with >64 stations, we'd have problems ...

I guess you can decide whether or not you want this patch then.

johannes
Jouni Malinen Jan. 21, 2020, 4:18 p.m. UTC | #2 
On Wed, Jan 15, 2020 at 12:27:20PM +0100, Johannes Berg wrote:
> On Wed, 2020-01-15 at 12:09 +0100, Johannes Berg wrote:
> > From: Johannes Berg <johannes.berg@intel.com>
> > 
> > These are all going to the kernel, and with powersave now
> > supported in hwsim, the frames are not transmitted immediately
> > but queued up. But the kernel has a limit of 64 queued up
> > frames when TX status ACK is requested, so this test fails.
> > Give it a minimal amount of time between frames so that they
> > can actually go out - 10ms for each means we get only 10 in
> > each beacon period before the other side wakes up and pulls
> > them, so we don't hit the 64 limit.
> 
> I'll fix the kernel too to increase that limit to 8192, since if this
> happens on an AP with >64 stations, we'd have problems ...
> 
> I guess you can decide whether or not you want this patch then.

Since it looks likely for the mac80211 change to get in, I'll drop this.

https://patchwork.kernel.org/patch/11333793/
diff mbox series

Patch

diff --git a/tests/hwsim/test_rrm.py b/tests/hwsim/test_rrm.py
index dde98932b6a6..34939e1c5129 100644
--- a/tests/hwsim/test_rrm.py
+++ b/tests/hwsim/test_rrm.py
@@ -1803,6 +1803,7 @@  def run_rrm_beacon_req_ap_errors(dev, apdev):
         raise Exception("Invalid REQ_BEACON accepted")
 
     for i in range(257):
+        time.sleep(0.01)
         if "FAIL" in hapd.request("REQ_BEACON " + addr + " 510b0000640000ffffffffffff"):
             raise Exception("REQ_BEACON failed")
         dev[0].dump_monitor()