debugd: fix ping |
|||||||||
Issue descriptionThe crosh ping command has two problems: 1. (Brian reported, I confirmed) at least on R59 in verified mode, "ping -c # <foo>" hangs after # pings. (Control-c unhangs it.) 2. Typing control-C in the shell stops the pings but also prints summary statistics. In crosh, it does not. Brian writes: SandboxedProcess::KillProcessGroup sends SIGKILL, while ping only provides summary output if it completes (e.g., with --count) or gets SIGINT. This may be a good onboarding bug?
,
May 25 2017
,
May 25 2017
we probably could make the killing more graceful ... have it send SIGINT first, then wait for it to die, and if it's still around, SIGKILL it.
,
Jun 1 2017
,
Jun 1 2017
Cannot repro problem 1 on M60. Working on problem 2.
,
Jun 1 2017
Yeah, I'm not sure Luigi translated all our data on problem 1 completely: he noted that it was working on ToT (at time of filing), but I noted the following tests on devices I had sitting around: Kevin, R59 beta channel, normal mode: hang Kevin, R60, 9515.0.0 builder test image, dev mode: hang Gru, R60, 9472.0.0 builder test image, dev mode: hang So presumably it got fixed in R60, since 9515.0.0. And still buggy on R59. Could be nice to figure out what fixed it, but it's probably not a priority at all.
,
Jun 1 2017
FYI, I recently updated net-misc/iputils (https://chromium-review.googlesource.com/c/516063/), which landed in 9604.0.0, which may have changed the behavior of #1.
,
Jun 2 2017
Eric also landed some fd related changes in debugd in R60
,
Jun 7 2017
Minijail recently gained the power to forward signals to its child. Waiting on a fix to that feature in 730682 which should make problem 2 easy to fix.
,
Jun 15 2017
,
Jun 15 2017
,
Jun 20 2017
,
Jul 5 2017
,
Jan 22 2018
|
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by briannorris@chromium.org
, May 25 2017