1317611 - Crash in mozilla::net::CrashWithReason

Status: RESOLVED DUPLICATE of bug 1315905

(Core :: Networking: HTTP, defect)

Description

[Ting-Yu Chou [:ting] (away)]

This bug was filed from the Socorro interface and is 
report bp-3804a7bc-4d7d-442b-91ca-020bf2161115.
=============================================================
#7 top crash on Windows for Nightly 52.0a1 20161113030203, 23 crashes from 19 installations. The crash reason is: 

  "GetValidatedAppInfo | App does not have permission"

Bug 1289001 marked 52 fixed, but it doesn't seem the case.

Comment 1

[Marco Castelluccio [:marco]]

(In reply to Ting-Yu Chou [:ting] from comment #0)
> Bug 1289001 marked 52 fixed, but it doesn't seem the case.

The signature is related to several different crashes (bug 1289001 fixed two different crashes), so it's better to open a new bug for the remaining crashes rather than fixing everything in a single bug.

Comment 2

[Andrea Marchesini [:baku]]

Attachment: necko_test.patch

I would like to see what the child process does when the channel is created.
With this patch I do:

1. sync HTTP Channel creation
2. the check happens in RecvPHttpChannelConstructor so that we can kill the child sending IPC_FAIL.

Comment 3

[Andrea Marchesini [:baku]]

Of course this patch should stay in m-i/m-c just for a couple of days. No more.

Comment 4

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

Comment on attachment 8813174 [details] [diff] [review]
necko_test.patch

Review of attachment 8813174 [details] [diff] [review]:
-----------------------------------------------------------------

I could be wrong, but I think that bug 1315905 eliminates this crash. UsingNeckoIPCSecurity is false on desktop, so we'll return here:
http://searchfox.org/mozilla-central/rev/59bb309e38b10aba63dea8505fb800e99fe821d6/netwerk/ipc/NeckoParent.cpp#219

Not sure if that was the intention. The patch would be fine if we were still crashing, so r+ :-). But please don't land this unless we resolve this issue, since it just adds a sync message pointlessly.

::: netwerk/ipc/NeckoParent.cpp
@@ +296,5 @@
> +  const char *error = CreateChannelLoadContext(aBrowser, Manager(),
> +                                               aSerialized, requestingPrincipal,
> +                                               loadContext);
> +  if (error) {
> +    printf_stderr("NeckoParent::AllocPHttpChannelParent: "

Did you mean to keep the printf around?

Comment 5

[Andrea Marchesini [:baku]]

Marco, can you help me to confirm (or not) that the bug has been already resolved by bug 1315905 ?

Comment 6

[Valentin Gosu [:valentin] (he/him)]

(In reply to Andrea Marchesini [:baku] from comment #5)
> Marco, can you help me to confirm (or not) that the bug has been already
> resolved by bug 1315905 ?

I see no crashes on 53.0a1 since bug 1315905 landed.

Comment 7

[Marco Castelluccio [:marco]]

Yes, the last build ID in the crash reports is 20161117030212, the first build ID containing the fix from bug 1315905 is 20161118030222.

So it looks like bug 1315905 fixed the crash (unless the signature changed again).

https://crash-stats.mozilla.com/search/?signature=%3Dmozilla%3A%3Anet%3A%3ACrashWithReason&product=Firefox&version=53.0a1&date=%3E%3D2016-11-16T12%3A09%3A07.000Z&date=%3C2016-11-23T12%3A09%3A07.000Z&_sort=-date&_facets=signature&_facets=build_id&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-build_id

Comment 8

[Andrea Marchesini [:baku]]

Valentin, let me know if you want to reopen it.

Comment 9

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

I filed bug 1319881 to figure out what to do here.