I've taken several patches from amdgpu-pro libdrm that look useful to me and I think can be applied already. The only things I did was rebasing, fixing some typos and dropping Change-Id.
Alex Xie (3): amdgpu: Free/uninit vamgr_32 in theoretically correct order amdgpu: vamgr can be a struct instead of a pointer amdgpu: vamgr_32 can be a struct instead of a pointer
amdgpu/amdgpu_device.c | 24 +++++++----------------- amdgpu/amdgpu_internal.h | 4 ++-- amdgpu/amdgpu_vamgr.c | 6 +++--- 3 files changed, 12 insertions(+), 22 deletions(-)
From: Alex Xie AlexBin.Xie@amd.com
vamgr_32 is a region inside general VAM range. It is better to free and deinitialize it before general VAM range.
Signed-off-by: Alex Xie AlexBin.Xie@amd.com Reviewed-by: Christian König christian.koenig@amd.com --- amdgpu/amdgpu_device.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c index f4ede03..cad7133 100644 --- a/amdgpu/amdgpu_device.c +++ b/amdgpu/amdgpu_device.c @@ -131,10 +131,10 @@ static int amdgpu_get_auth(int fd, int *auth)
static void amdgpu_device_free_internal(amdgpu_device_handle dev) { - amdgpu_vamgr_deinit(dev->vamgr); - free(dev->vamgr); amdgpu_vamgr_deinit(dev->vamgr_32); free(dev->vamgr_32); + amdgpu_vamgr_deinit(dev->vamgr); + free(dev->vamgr); util_hash_table_destroy(dev->bo_flink_names); util_hash_table_destroy(dev->bo_handles); pthread_mutex_destroy(&dev->bo_table_mutex);
From: Alex Xie AlexBin.Xie@amd.com
vamgr_32 is an integral part of amdgpu_device. We don't need to calloc and free it. This can save CPU time, reduce heap fragmentation.
Signed-off-by: Alex Xie AlexBin.Xie@amd.com Reviewed-by: Christian König christian.koenig@amd.com [Grazvydas Ignotas: rebase, correct a typo in commit message] --- amdgpu/amdgpu_device.c | 8 ++------ amdgpu/amdgpu_internal.h | 2 +- amdgpu/amdgpu_vamgr.c | 4 ++-- 3 files changed, 5 insertions(+), 9 deletions(-)
diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c index cad7133..11714e4 100644 --- a/amdgpu/amdgpu_device.c +++ b/amdgpu/amdgpu_device.c @@ -131,8 +131,7 @@ static int amdgpu_get_auth(int fd, int *auth)
static void amdgpu_device_free_internal(amdgpu_device_handle dev) { - amdgpu_vamgr_deinit(dev->vamgr_32); - free(dev->vamgr_32); + amdgpu_vamgr_deinit(&dev->vamgr_32); amdgpu_vamgr_deinit(dev->vamgr); free(dev->vamgr); util_hash_table_destroy(dev->bo_flink_names); @@ -270,10 +269,7 @@ int amdgpu_device_initialize(int fd, if (start > 0xffffffff) goto free_va; /* shouldn't get here */
- dev->vamgr_32 = calloc(1, sizeof(struct amdgpu_bo_va_mgr)); - if (dev->vamgr_32 == NULL) - goto free_va; - amdgpu_vamgr_init(dev->vamgr_32, start, max, + amdgpu_vamgr_init(&dev->vamgr_32, start, max, dev->dev_info.virtual_address_alignment);
*major_version = dev->major_version; diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h index 4f039b6..7e237ac 100644 --- a/amdgpu/amdgpu_internal.h +++ b/amdgpu/amdgpu_internal.h @@ -87,7 +87,7 @@ struct amdgpu_device { /** The global VA manager for the whole virtual address space */ struct amdgpu_bo_va_mgr *vamgr; /** The VA manager for the 32bit address space */ - struct amdgpu_bo_va_mgr *vamgr_32; + struct amdgpu_bo_va_mgr vamgr_32; };
struct amdgpu_bo { diff --git a/amdgpu/amdgpu_vamgr.c b/amdgpu/amdgpu_vamgr.c index 8a707cb..4dc4253 100644 --- a/amdgpu/amdgpu_vamgr.c +++ b/amdgpu/amdgpu_vamgr.c @@ -236,7 +236,7 @@ int amdgpu_va_range_alloc(amdgpu_device_handle dev, struct amdgpu_bo_va_mgr *vamgr;
if (flags & AMDGPU_VA_RANGE_32_BIT) - vamgr = dev->vamgr_32; + vamgr = &dev->vamgr_32; else vamgr = dev->vamgr;
@@ -249,7 +249,7 @@ int amdgpu_va_range_alloc(amdgpu_device_handle dev, if (!(flags & AMDGPU_VA_RANGE_32_BIT) && (*va_base_allocated == AMDGPU_INVALID_VA_ADDRESS)) { /* fallback to 32bit address */ - vamgr = dev->vamgr_32; + vamgr = &dev->vamgr_32; *va_base_allocated = amdgpu_vamgr_find_va(vamgr, size, va_base_alignment, va_base_required); }
This series is, Reviewed-by: Edward O'Callaghan funfunctor@folklore1984.net
On 01/29/2017 06:49 AM, Grazvydas Ignotas wrote:
I've taken several patches from amdgpu-pro libdrm that look useful to me and I think can be applied already. The only things I did was rebasing, fixing some typos and dropping Change-Id.
Alex Xie (3): amdgpu: Free/uninit vamgr_32 in theoretically correct order amdgpu: vamgr can be a struct instead of a pointer amdgpu: vamgr_32 can be a struct instead of a pointer
amdgpu/amdgpu_device.c | 24 +++++++----------------- amdgpu/amdgpu_internal.h | 4 ++-- amdgpu/amdgpu_vamgr.c | 6 +++--- 3 files changed, 12 insertions(+), 22 deletions(-)
Reviewed-by: Christian König christian.koenig@amd.com.
Am 29.01.2017 um 23:51 schrieb Edward O'Callaghan:
This series is, Reviewed-by: Edward O'Callaghan funfunctor@folklore1984.net
On 01/29/2017 06:49 AM, Grazvydas Ignotas wrote:
I've taken several patches from amdgpu-pro libdrm that look useful to me and I think can be applied already. The only things I did was rebasing, fixing some typos and dropping Change-Id.
Alex Xie (3): amdgpu: Free/uninit vamgr_32 in theoretically correct order amdgpu: vamgr can be a struct instead of a pointer amdgpu: vamgr_32 can be a struct instead of a pointer
amdgpu/amdgpu_device.c | 24 +++++++----------------- amdgpu/amdgpu_internal.h | 4 ++-- amdgpu/amdgpu_vamgr.c | 6 +++--- 3 files changed, 12 insertions(+), 22 deletions(-)
amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx
On 28 January 2017 at 19:49, Grazvydas Ignotas notasas@gmail.com wrote:
I've taken several patches from amdgpu-pro libdrm that look useful to me and I think can be applied already. The only things I did was rebasing, fixing some typos and dropping Change-Id.
Alex Xie (3): amdgpu: Free/uninit vamgr_32 in theoretically correct order amdgpu: vamgr can be a struct instead of a pointer amdgpu: vamgr_32 can be a struct instead of a pointer
Nice cleanups/fixes indeed.
Fwiw Reviewed-by: Emil Velikov emil.l.velikov@gmail.com
Unless someone beats me to it I'll collect all the tags and push these over the weekend. Emil
dri-devel@lists.freedesktop.org
-
Christian König -
Edward O'Callaghan -
Emil Velikov -
Grazvydas Ignotas