Proxmox-Port/linux-loongson
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson synced 2025-09-06 13:09:58 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

drm/amdkfd: enable single alu ops for gfx12

Browse Source 
GFX12 debugging requires setting up precise ALU operation for catching
ALU exceptions.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Tested-by: Lancelot Six <lancelot.six@amd.com>
Reviewed-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Jonathan Kim 2023-08-21 11:47:47 -04:00 committed by Alex Deucher
parent 984b265ff6
commit 9243240bed
4 changed files with 28 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
drivers/gpu/drm/amd/amdkfd/kfd_debug.c
View File

@ -25,6 +25,7 @@
#include "kfd_topology.h"
#include <linux/file.h>
#include <uapi/linux/kfd_ioctl.h>
#include <uapi/linux/kfd_sysfs.h>
#define MAX_WATCH_ADDRESSES 4
@ -497,14 +498,24 @@ int kfd_dbg_trap_set_flags(struct kfd_process *target, uint32_t *flags)
int i, r = 0, rewind_count = 0;
for (i = 0; i < target->n_pdds; i++) {
if (!kfd_dbg_is_per_vmid_supported(target->pdds[i]->dev) &&
struct kfd_topology_device *topo_dev =
kfd_topology_device_by_id(target->pdds[i]->dev->id);
uint32_t caps = topo_dev->node_props.capability;
if (!(caps | HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED) &&
(*flags & KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP)) {
*flags = prev_flags;
return -EACCES;
}
if (!(caps | HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED) &&
(*flags & KFD_DBG_TRAP_FLAG_SINGLE_ALU_OP)) {
*flags = prev_flags;
return -EACCES;
}
}
target->dbg_flags = *flags & KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP;
target->dbg_flags = *flags;
*flags = prev_flags;
for (i = 0; i < target->n_pdds; i++) {
struct kfd_process_device *pdd = target->pdds[i];

4
drivers/gpu/drm/amd/amdkfd/kfd_topology.c
View File

@ -1929,6 +1929,10 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0))
dev->node_props.capability |=
HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0))
dev->node_props.capability |=
HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED;
}
kfd_topology_set_dbg_firmware_support(dev);

1
include/uapi/linux/kfd_ioctl.h
View File

@ -854,6 +854,7 @@ enum kfd_dbg_trap_address_watch_mode {
/* Additional wave settings */
enum kfd_dbg_trap_flags {
KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP = 1,
KFD_DBG_TRAP_FLAG_SINGLE_ALU_OP = 2,
};
/* Trap exceptions */

19
include/uapi/linux/kfd_sysfs.h
View File

@ -51,15 +51,16 @@
/* Old buggy user mode depends on this being 0 */
#define HSA_CAP_RESERVED_WAS_SRAM_EDCSUPPORTED 0x00080000
#define HSA_CAP_MEM_EDCSUPPORTED 0x00100000
#define HSA_CAP_RASEVENTNOTIFY 0x00200000
#define HSA_CAP_ASIC_REVISION_MASK 0x03c00000
#define HSA_CAP_ASIC_REVISION_SHIFT 22
#define HSA_CAP_SRAM_EDCSUPPORTED 0x04000000
#define HSA_CAP_SVMAPI_SUPPORTED 0x08000000
#define HSA_CAP_FLAGS_COHERENTHOSTACCESS 0x10000000
#define HSA_CAP_TRAP_DEBUG_FIRMWARE_SUPPORTED 0x20000000
#define HSA_CAP_RESERVED 0xe00f8000
#define HSA_CAP_MEM_EDCSUPPORTED 0x00100000
#define HSA_CAP_RASEVENTNOTIFY 0x00200000
#define HSA_CAP_ASIC_REVISION_MASK 0x03c00000
#define HSA_CAP_ASIC_REVISION_SHIFT 22
#define HSA_CAP_SRAM_EDCSUPPORTED 0x04000000
#define HSA_CAP_SVMAPI_SUPPORTED 0x08000000
#define HSA_CAP_FLAGS_COHERENTHOSTACCESS 0x10000000
#define HSA_CAP_TRAP_DEBUG_FIRMWARE_SUPPORTED 0x20000000
#define HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED 0x40000000
#define HSA_CAP_RESERVED 0x800f8000
/* debug_prop bits in node properties */
#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_MASK 0x0000000f