On Tue, Nov 22, 2016 at 2:31 PM, Emil Velikov <[email protected]> wrote: > On 22 November 2016 at 01:19, Bas Nieuwenhuizen <[email protected]> > wrote: >> We want to also invalidate the cache when LLVM gets changed. As the >> specific LLVM revision is not fixed at build time, we will need to >> check at runtime. Computing a checksum for LLVM is going to be very >> expensive, so just use the mtime. > Good catch Bas ! Also - woohoo, no more RADV_TIMESTAMP. > >> +static int >> radv_device_get_cache_uuid(void *uuid) >> { >> + uint32_t mesa_timestamp, llvm_timestamp; >> memset(uuid, 0, VK_UUID_SIZE); >> - snprintf(uuid, VK_UUID_SIZE, "radv-%s", RADV_TIMESTAMP); >> + if (radv_get_function_timestamp(radv_device_get_cache_uuid, >> &mesa_timestamp) || >> + radv_get_function_timestamp(LLVMInitializeAMDGPUTargetInfo, >> &llvm_timestamp)) > Can you please double-check that this works correctly with static LLVM ?
Done. It just points to the libvulkan_radeon.so then and becomes a redundant timestamp. - Bas > >> + return -1; >> + >> + memcpy(uuid, &mesa_timestamp, 4); >> + memcpy((char*)uuid + 4, &llvm_timestamp, 4); >> + snprintf((char*)uuid + 8, VK_UUID_SIZE - 8, "radv"); > Since radv_timestamp.h and RADV_TIMESTAMP are no longer used can I ask > you to remove the machinery which sets up ? And/Or update ANV? > With follow-up patch(es) of course. > > Thanks > Emil _______________________________________________ mesa-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/mesa-dev
