[Toybox] Release prep.

Andreas Gampe agampe at google.com
Wed Feb 7 08:41:22 PST 2018


Stack Trace:
  RELADDR           FUNCTION         FILE:LINE
     7e253  top_common+6387  external/toybox/toys/posix/ps.c:1420
     7c413  top_main+555     external/toybox/toys/posix/ps.c:1666
     1f7db  toy_exec+311     external/toybox/main.c:169
     1ef77  toybox_main+91   external/toybox/main.c:182
     1fa27  main+431         external/toybox/main.c:240
     ac8a3  __libc_init+91   bionic/libc/bionic/libc_init_dynamic.cpp:129

ps.c:1419f

    while (old.count || new.count) {
      struct carveup *otb = *old.tb, *ntb = *new.tb;

You probably should not proactively dereference both old.tb and
new.tb. One might be invalid if not both old.count and new.count > 0.


Best,
 Andreas



On Wed, Feb 7, 2018 at 8:35 AM, enh <enh at google.com> wrote:
> On Mon, Feb 5, 2018 at 11:01 AM, enh <enh at google.com> wrote:
>> On Mon, Feb 5, 2018 at 10:43 AM, Rob Landley <rob at landley.net> wrote:
>>> On 02/05/2018 11:34 AM, enh wrote:
>>>> On Sun, Feb 4, 2018 at 6:14 PM, Rob Landley <rob at landley.net> wrote:
>>>>> For some reason I thought the next release should happen the middle of
>>>>> this month, but it turns out it was the middle of last month. (Ok, I've
>>>>> been distracted.)
>>>>>
>>>>> So I'm giving myself one more week to tidy up loose ends. I need to fix
>>>>> ps -T, might get ping ready to promote...
>>>>>
>>>>> Anybody want to shout out with something important I forgot?
>>>>
>>>> the asan failures in top are being seen more and more often as folks
>>>> (internal and OEMs) run asan builds on a regular basis. (moore's law
>>>> ftw!)
>>>
>>> This is why I spent so much effort wrangling the NDK, let's see if I can
>>> reproduce it...
>>
>> i don't think you need to. i'm pretty sure i've seen it on the host
>> too. (the problem, like i said, is that for me it takes a long time to
>> reproduce by just leaving top running, even though i'm sure that the
>> crashes that get automatically reported aren't coming from long-lived
>> top instances.)
>>
>> yeah, checking my archive, i see that i reported reproducing this on
>> the host in the thread surrounding "[PATCH] killall should kill
>> scripts too".
>
> +agampe who seems to be hitting this more frequently now and can
> supply a current detailed asan abort...
>
>>>   error: cannot specify -static with -fsanitize=address.
>>>
>>> errorier: cannot run bionic binary on ubuntu host without -static.
>>> (Sigh, I'll see if I can set up a bionic chroot this evening...)
>>>
>>> Rob


More information about the Toybox mailing list