Quoting Joonas Lahtinen (2017-12-04 10:23:11)
> On Fri, 2017-12-01 at 11:13 +, Chris Wilson wrote:
> > +++ b/tests/gem_exec_fence.c
> > @@ -698,7 +698,7 @@ static bool has_submit_fence(int fd)
> > int value = 0;
> >
> > memset(, 0, sizeof(gp));
> > - gp.param = 50; /*
On Fri, 2017-12-01 at 11:13 +, Chris Wilson wrote:
> A new context assumes that all of its registers are in the default state
> when it is created. What may happen is that a register written by one
> context may leak into the second, causing mass confusion.
>
> v2: Extend back to Sandybridge
A new context assumes that all of its registers are in the default state
when it is created. What may happen is that a register written by one
context may leak into the second, causing mass confusion.
v2: Extend back to Sandybridge (etc)
v3: Check context preserves registers across
A new context assumes that all of its registers are in the default state
when it is created. What may happen is that a register written by one
context may leak into the second, causing mass confusion.
v2: Extend back to Sandybridge (etc)
v3: Check context preserves registers across
Quoting Chris Wilson (2017-10-24 12:07:58)
> A new context assumes that all of its registers are in the default state
> when it is created. What may happen is that a register written by one
> context may leak into the second, causing mass confusion.
>
> Signed-off-by: Chris Wilson
A new context assumes that all of its registers are in the default state
when it is created. What may happen is that a register written by one
context may leak into the second, causing mass confusion.
Signed-off-by: Chris Wilson
---
tests/Makefile.sources| 1 +