drm/i915/gt: remove some limited use register access wrappers
authorJani Nikula <jani.nikula@intel.com>
Wed, 23 Nov 2022 16:49:16 +0000 (18:49 +0200)
committerJani Nikula <jani.nikula@intel.com>
Wed, 7 Dec 2022 11:24:15 +0000 (13:24 +0200)
Remove rmw_set(), rmw_clear(), clear_register(), rmw_set_fw(), and
rmw_clear_fw(). They're just one too many levels of abstraction for
register access, for very specific purposes.

clear_register() seems like a micro-optimization bypassing the write
when the register is already clear, but that trick has ceased to work
since commit 06b975d58fd6 ("drm/i915: make intel_uncore_rmw() write
unconditionally"). Just clear the register in the most obvious way.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221123164916.4128733-1-jani.nikula@intel.com
drivers/gpu/drm/i915/gt/intel_gt.c
drivers/gpu/drm/i915/gt/intel_reset.c

index 42b9e2b..2bf2ec7 100644 (file)
@@ -218,21 +218,6 @@ out:
        return ret;
 }
 
-static void rmw_set(struct intel_uncore *uncore, i915_reg_t reg, u32 set)
-{
-       intel_uncore_rmw(uncore, reg, 0, set);
-}
-
-static void rmw_clear(struct intel_uncore *uncore, i915_reg_t reg, u32 clr)
-{
-       intel_uncore_rmw(uncore, reg, clr, 0);
-}
-
-static void clear_register(struct intel_uncore *uncore, i915_reg_t reg)
-{
-       intel_uncore_rmw(uncore, reg, 0, 0);
-}
-
 static void gen6_clear_engine_error_register(struct intel_engine_cs *engine)
 {
        GEN6_RING_FAULT_REG_RMW(engine, RING_FAULT_VALID, 0);
@@ -258,14 +243,14 @@ intel_gt_clear_error_registers(struct intel_gt *gt,
        u32 eir;
 
        if (GRAPHICS_VER(i915) != 2)
-               clear_register(uncore, PGTBL_ER);
+               intel_uncore_write(uncore, PGTBL_ER, 0);
 
        if (GRAPHICS_VER(i915) < 4)
-               clear_register(uncore, IPEIR(RENDER_RING_BASE));
+               intel_uncore_write(uncore, IPEIR(RENDER_RING_BASE), 0);
        else
-               clear_register(uncore, IPEIR_I965);
+               intel_uncore_write(uncore, IPEIR_I965, 0);
 
-       clear_register(uncore, EIR);
+       intel_uncore_write(uncore, EIR, 0);
        eir = intel_uncore_read(uncore, EIR);
        if (eir) {
                /*
@@ -273,7 +258,7 @@ intel_gt_clear_error_registers(struct intel_gt *gt,
                 * mask them.
                 */
                drm_dbg(&gt->i915->drm, "EIR stuck: 0x%08x, masking\n", eir);
-               rmw_set(uncore, EMR, eir);
+               intel_uncore_rmw(uncore, EMR, 0, eir);
                intel_uncore_write(uncore, GEN2_IIR,
                                   I915_MASTER_ERROR_INTERRUPT);
        }
@@ -283,10 +268,10 @@ intel_gt_clear_error_registers(struct intel_gt *gt,
                                           RING_FAULT_VALID, 0);
                intel_gt_mcr_read_any(gt, XEHP_RING_FAULT_REG);
        } else if (GRAPHICS_VER(i915) >= 12) {
-               rmw_clear(uncore, GEN12_RING_FAULT_REG, RING_FAULT_VALID);
+               intel_uncore_rmw(uncore, GEN12_RING_FAULT_REG, RING_FAULT_VALID, 0);
                intel_uncore_posting_read(uncore, GEN12_RING_FAULT_REG);
        } else if (GRAPHICS_VER(i915) >= 8) {
-               rmw_clear(uncore, GEN8_RING_FAULT_REG, RING_FAULT_VALID);
+               intel_uncore_rmw(uncore, GEN8_RING_FAULT_REG, RING_FAULT_VALID, 0);
                intel_uncore_posting_read(uncore, GEN8_RING_FAULT_REG);
        } else if (GRAPHICS_VER(i915) >= 6) {
                struct intel_engine_cs *engine;
index 24736eb..ffde89c 100644 (file)
 /* XXX How to handle concurrent GGTT updates using tiling registers? */
 #define RESET_UNDER_STOP_MACHINE 0
 
-static void rmw_set_fw(struct intel_uncore *uncore, i915_reg_t reg, u32 set)
-{
-       intel_uncore_rmw_fw(uncore, reg, 0, set);
-}
-
-static void rmw_clear_fw(struct intel_uncore *uncore, i915_reg_t reg, u32 clr)
-{
-       intel_uncore_rmw_fw(uncore, reg, clr, 0);
-}
-
 static void client_mark_guilty(struct i915_gem_context *ctx, bool banned)
 {
        struct drm_i915_file_private *file_priv = ctx->file_priv;
@@ -212,7 +202,7 @@ static int g4x_do_reset(struct intel_gt *gt,
        int ret;
 
        /* WaVcpClkGateDisableForMediaReset:ctg,elk */
-       rmw_set_fw(uncore, VDECCLK_GATE_D, VCP_UNIT_CLOCK_GATE_DISABLE);
+       intel_uncore_rmw_fw(uncore, VDECCLK_GATE_D, 0, VCP_UNIT_CLOCK_GATE_DISABLE);
        intel_uncore_posting_read_fw(uncore, VDECCLK_GATE_D);
 
        pci_write_config_byte(pdev, I915_GDRST,
@@ -234,7 +224,7 @@ static int g4x_do_reset(struct intel_gt *gt,
 out:
        pci_write_config_byte(pdev, I915_GDRST, 0);
 
-       rmw_clear_fw(uncore, VDECCLK_GATE_D, VCP_UNIT_CLOCK_GATE_DISABLE);
+       intel_uncore_rmw_fw(uncore, VDECCLK_GATE_D, VCP_UNIT_CLOCK_GATE_DISABLE, 0);
        intel_uncore_posting_read_fw(uncore, VDECCLK_GATE_D);
 
        return ret;
@@ -448,7 +438,7 @@ static int gen11_lock_sfc(struct intel_engine_cs *engine,
         * to reset it as well (we will unlock it once the reset sequence is
         * completed).
         */
-       rmw_set_fw(uncore, sfc_lock.lock_reg, sfc_lock.lock_bit);
+       intel_uncore_rmw_fw(uncore, sfc_lock.lock_reg, 0, sfc_lock.lock_bit);
 
        ret = __intel_wait_for_register_fw(uncore,
                                           sfc_lock.ack_reg,
@@ -498,7 +488,7 @@ static void gen11_unlock_sfc(struct intel_engine_cs *engine)
 
        get_sfc_forced_lock_data(engine, &sfc_lock);
 
-       rmw_clear_fw(uncore, sfc_lock.lock_reg, sfc_lock.lock_bit);
+       intel_uncore_rmw_fw(uncore, sfc_lock.lock_reg, sfc_lock.lock_bit, 0);
 }
 
 static int __gen11_reset_engines(struct intel_gt *gt,