[Intel-gfx] [RFC PATCH i-g-t] lib/i915: Restrict mmap types to GTT if no MMAP_OFFSET support
Chris Wilson
chris at chris-wilson.co.uk
Thu Feb 20 16:09:14 UTC 2020
Quoting Janusz Krzysztofik (2020-02-20 15:32:03)
> Commit b0da8bb705c0 ("lib/i915: for_each_mmap_offset_type()")
> introduced a macro that makes it easy to repeat a test body within a
> loop for each mmap-offset mapping type supported by v4 of i915 MMAP_GTT
> API. However, when run on an older version of the driver, those
> subtests are believed to be still repeated for each known mmap-offset
> mapping type while effectively exercising GTT mapping type only. As
> that may be confusing, fix it.
>
> It has been assumed that the modified macro is still suitable for use
> inside gem_mmap_offset test itself. Would that not be case,
> gem_mmap_offset could redefine the macro back to its initial form for
> internal use.
>
> Suggested-by: Michał Winiarski <michal.winiarski at intel.com>
> Signed-off-by: Janusz Krzysztofik <janusz.krzysztofik at linux.intel.com>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> ---
> lib/i915/gem_mman.h | 5 +++--
> tests/i915/gem_ctx_sseu.c | 2 +-
> tests/i915/gem_exec_params.c | 2 +-
> tests/i915/gem_madvise.c | 18 ++++++++++++++----
> tests/i915/gem_mmap_offset.c | 10 +++++-----
> tests/i915/i915_pm_rpm.c | 2 +-
> 6 files changed, 25 insertions(+), 14 deletions(-)
>
> diff --git a/lib/i915/gem_mman.h b/lib/i915/gem_mman.h
> index 4fc6a0186..491767023 100644
> --- a/lib/i915/gem_mman.h
> +++ b/lib/i915/gem_mman.h
> @@ -101,9 +101,10 @@ extern const struct mmap_offset {
> unsigned int domain;
> } mmap_offset_types[];
>
> -#define for_each_mmap_offset_type(__t) \
> +#define for_each_mmap_offset_type(fd, __t) \
> for (const struct mmap_offset *__t = mmap_offset_types; \
> - (__t)->name; \
> + (__t)->name && (gem_has_mmap_offset(fd) || \
> + (__t)->type == I915_MMAP_OFFSET_GTT); \
Sigh.
extern bool gem_has_mmap_offset_type(int i915, const struct mmap_offset *t);
&& gem_has_mmap_offset_type(fd, t)
in case we need to fix it again in future.
-Chris
More information about the Intel-gfx
mailing list