Re: [PATCH weston] text_backend: make destructor call explicit

2015-06-25 Thread Pekka Paalanen
On Thu, 25 Jun 2015 09:06:52 -0500 Derek Foreman wrote: > Oops, I spoke too soon... > > On 25/06/15 09:01 AM, Derek Foreman wrote: > > On 24/06/15 08:28 AM, Pekka Paalanen wrote: > >> From: Pekka Paalanen > >> > >> We used to rely on the order in which the > >> weston_compositor::destroy_signal

Re: [PATCH weston] text_backend: make destructor call explicit

2015-06-25 Thread Derek Foreman
On 24/06/15 08:28 AM, Pekka Paalanen wrote: > From: Pekka Paalanen > > We used to rely on the order in which the > weston_compositor::destroy_signal callbacks happened, to not access > freed memory. Don't know when, but this broke at least with ivi-shell, > which caused crashes in random places o

Re: [PATCH weston] text_backend: make destructor call explicit

2015-06-25 Thread Derek Foreman
Oops, I spoke too soon... On 25/06/15 09:01 AM, Derek Foreman wrote: > On 24/06/15 08:28 AM, Pekka Paalanen wrote: >> From: Pekka Paalanen >> >> We used to rely on the order in which the >> weston_compositor::destroy_signal callbacks happened, to not access >> freed memory. Don't know when, but t

[PATCH weston] text_backend: make destructor call explicit

2015-06-24 Thread Pekka Paalanen
From: Pekka Paalanen We used to rely on the order in which the weston_compositor::destroy_signal callbacks happened, to not access freed memory. Don't know when, but this broke at least with ivi-shell, which caused crashes in random places on compositor shutdown. Valgrind found the following: