Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 4FCDBEB64D7 for ; Tue, 20 Jun 2023 19:35:10 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 24DDD10E383; Tue, 20 Jun 2023 19:35:10 +0000 (UTC) Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by gabe.freedesktop.org (Postfix) with ESMTPS id E8BF610E37F for ; Tue, 20 Jun 2023 19:35:05 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1687289705; x=1718825705; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=1BoWcT9pM6nwBb8cIPdIG0IWnDt3oibA9haEHG08K24=; b=nERD9V1GNyzYEmd8CQ6jGUZIS2HW/Q0UKImlkc723UHCBoWhOeo2iTbv YNiurrj36nJRfpw06bHO31cgqhj8xGa6vWQxmX8oivFEIo/euPtwl9bpT 5D88UhJ7WqjwPs0tzYIC+UADe6kkV4riGnc/dmUHtgsjdib+APFU17Rak YdYIiPH1QWCnL6pQYKFyGzc+SjUApUUjK/D4qzYOLfcYegy5nOgQlThrC Ggy3PAJUVIZNzAHn7ggBSDQH6fFHQSOnkWGAKXifmm7yBLa7CGIQVyZuE 8L3mUCTVDpQhvHcBjkwxnGOcF5yGm3kTO9Zs5Tb3MDl/YeM0XnKEe0pgZ Q==; X-IronPort-AV: E=McAfee;i="6600,9927,10747"; a="339579531" X-IronPort-AV: E=Sophos;i="6.00,258,1681196400"; d="scan'208";a="339579531" Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 20 Jun 2023 12:35:05 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10747"; a="858706277" X-IronPort-AV: E=Sophos;i="6.00,258,1681196400"; d="scan'208";a="858706277" Received: from invictus.jf.intel.com ([10.165.21.201]) by fmsmga001-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 20 Jun 2023 12:35:04 -0700 From: Radhakrishna Sripada To: intel-xe@lists.freedesktop.org Date: Tue, 20 Jun 2023 12:32:29 -0700 Message-Id: <20230620193229.1698372-3-radhakrishna.sripada@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20230620193229.1698372-1-radhakrishna.sripada@intel.com> References: <20230620193229.1698372-1-radhakrishna.sripada@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [Intel-xe] [PATCH v2 2/2] drm/xe: Add reg read/write trace X-BeenThere: intel-xe@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel Xe graphics driver List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Matt Roper , Lucas De Marchi Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" This will help debug register read/writes and provide a way to trace all the mmio transactions. In order to avoid spam from xe_mmio_wait32, xe_mmio_read32_notrace is introduced and used. v2: Do not use inline for the function declarations.(MattR) Cc: Lucas De Marchi Cc: Matt Roper Signed-off-by: Radhakrishna Sripada --- .../drm/xe/compat-i915-headers/intel_uncore.h | 8 +- drivers/gpu/drm/xe/xe_mmio.c | 80 ++++++++++++++++++- drivers/gpu/drm/xe/xe_mmio.h | 5 ++ drivers/gpu/drm/xe/xe_trace.h | 26 ++++++ 4 files changed, 111 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h b/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h index fae6213d26f1..036f3b963a84 100644 --- a/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h +++ b/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h @@ -113,7 +113,7 @@ static inline u32 intel_uncore_read_fw(struct fake_uncore *uncore, { struct xe_reg reg = XE_REG(i915_mmio_reg_offset(i915_reg)); - return xe_mmio_read32(__fake_uncore_to_gt(uncore), reg); + return xe_mmio_read32_notrace(__fake_uncore_to_gt(uncore), reg); } static inline void intel_uncore_write_fw(struct fake_uncore *uncore, @@ -121,7 +121,7 @@ static inline void intel_uncore_write_fw(struct fake_uncore *uncore, { struct xe_reg reg = XE_REG(i915_mmio_reg_offset(i915_reg)); - xe_mmio_write32(__fake_uncore_to_gt(uncore), reg, val); + xe_mmio_write32_notrace(__fake_uncore_to_gt(uncore), reg, val); } static inline u32 intel_uncore_read_notrace(struct fake_uncore *uncore, @@ -129,7 +129,7 @@ static inline u32 intel_uncore_read_notrace(struct fake_uncore *uncore, { struct xe_reg reg = XE_REG(i915_mmio_reg_offset(i915_reg)); - return xe_mmio_read32(__fake_uncore_to_gt(uncore), reg); + return xe_mmio_read32_notrace(__fake_uncore_to_gt(uncore), reg); } static inline void intel_uncore_write_notrace(struct fake_uncore *uncore, @@ -137,7 +137,7 @@ static inline void intel_uncore_write_notrace(struct fake_uncore *uncore, { struct xe_reg reg = XE_REG(i915_mmio_reg_offset(i915_reg)); - xe_mmio_write32(__fake_uncore_to_gt(uncore), reg, val); + xe_mmio_write32_notrace(__fake_uncore_to_gt(uncore), reg, val); } #endif /* __INTEL_UNCORE_H__ */ diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c index 78c7f839f783..ef2192a2d060 100644 --- a/drivers/gpu/drm/xe/xe_mmio.c +++ b/drivers/gpu/drm/xe/xe_mmio.c @@ -19,6 +19,7 @@ #include "xe_gt_mcr.h" #include "xe_macros.h" #include "xe_module.h" +#include "xe_trace.h" #define XEHP_MTCFG_ADDR XE_REG(0x101800) #define TILE_COUNT REG_GENMASK(15, 8) @@ -429,7 +430,7 @@ static const struct xe_reg mmio_read_whitelist[] = { RING_TIMESTAMP(RENDER_RING_BASE), }; -u8 xe_mmio_read8(struct xe_gt *gt, struct xe_reg reg) +u8 xe_mmio_read8_notrace(struct xe_gt *gt, struct xe_reg reg) { struct xe_tile *tile = gt_to_tile(gt); @@ -439,6 +440,31 @@ u8 xe_mmio_read8(struct xe_gt *gt, struct xe_reg reg) return readb(tile->mmio.regs + reg.addr); } +u8 xe_mmio_read8(struct xe_gt *gt, struct xe_reg reg) +{ + struct xe_tile *tile = gt_to_tile(gt); + u8 val; + + if (reg.addr < gt->mmio.adj_limit) + reg.addr += gt->mmio.adj_offset; + + val = readb(tile->mmio.regs + reg.addr); + + trace_xe_reg_rw(false, ®, val, sizeof(val)); + + return val; +} + +void xe_mmio_write32_notrace(struct xe_gt *gt, struct xe_reg reg, u32 val) +{ + struct xe_tile *tile = gt_to_tile(gt); + + if (reg.addr < gt->mmio.adj_limit) + reg.addr += gt->mmio.adj_offset; + + writel(val, tile->mmio.regs + reg.addr); +} + void xe_mmio_write32(struct xe_gt *gt, struct xe_reg reg, u32 val) { struct xe_tile *tile = gt_to_tile(gt); @@ -446,10 +472,12 @@ void xe_mmio_write32(struct xe_gt *gt, struct xe_reg reg, u32 val) if (reg.addr < gt->mmio.adj_limit) reg.addr += gt->mmio.adj_offset; + trace_xe_reg_rw(true, ®, val, sizeof(val)); + writel(val, tile->mmio.regs + reg.addr); } -u32 xe_mmio_read32(struct xe_gt *gt, struct xe_reg reg) +u32 xe_mmio_read32_notrace(struct xe_gt *gt, struct xe_reg reg) { struct xe_tile *tile = gt_to_tile(gt); @@ -459,6 +487,21 @@ u32 xe_mmio_read32(struct xe_gt *gt, struct xe_reg reg) return readl(tile->mmio.regs + reg.addr); } +u32 xe_mmio_read32(struct xe_gt *gt, struct xe_reg reg) +{ + struct xe_tile *tile = gt_to_tile(gt); + u32 val; + + if (reg.addr < gt->mmio.adj_limit) + reg.addr += gt->mmio.adj_offset; + + val = readl(tile->mmio.regs + reg.addr); + + trace_xe_reg_rw(false, ®, val, sizeof(val)); + + return val; +} + u32 xe_mmio_rmw32(struct xe_gt *gt, struct xe_reg reg, u32 clr, u32 set) { u32 old, reg_val; @@ -470,6 +513,16 @@ u32 xe_mmio_rmw32(struct xe_gt *gt, struct xe_reg reg, u32 clr, u32 set) return old; } +void xe_mmio_write64_notrace(struct xe_gt *gt, struct xe_reg reg, u64 val) +{ + struct xe_tile *tile = gt_to_tile(gt); + + if (reg.addr < gt->mmio.adj_limit) + reg.addr += gt->mmio.adj_offset; + + writeq(val, tile->mmio.regs + reg.addr); +} + void xe_mmio_write64(struct xe_gt *gt, struct xe_reg reg, u64 val) { struct xe_tile *tile = gt_to_tile(gt); @@ -477,10 +530,12 @@ void xe_mmio_write64(struct xe_gt *gt, struct xe_reg reg, u64 val) if (reg.addr < gt->mmio.adj_limit) reg.addr += gt->mmio.adj_offset; + trace_xe_reg_rw(true, ®, val, sizeof(val)); + writeq(val, tile->mmio.regs + reg.addr); } -u64 xe_mmio_read64(struct xe_gt *gt, struct xe_reg reg) +u64 xe_mmio_read64_notrace(struct xe_gt *gt, struct xe_reg reg) { struct xe_tile *tile = gt_to_tile(gt); @@ -490,6 +545,21 @@ u64 xe_mmio_read64(struct xe_gt *gt, struct xe_reg reg) return readq(tile->mmio.regs + reg.addr); } +u64 xe_mmio_read64(struct xe_gt *gt, struct xe_reg reg) +{ + struct xe_tile *tile = gt_to_tile(gt); + u64 val; + + if (reg.addr < gt->mmio.adj_limit) + reg.addr += gt->mmio.adj_offset; + + val = readq(tile->mmio.regs + reg.addr); + + trace_xe_reg_rw(false, ®, val, sizeof(val)); + + return val; +} + int xe_mmio_write32_and_verify(struct xe_gt *gt, struct xe_reg reg, u32 val, u32 mask, u32 eval) { @@ -511,7 +581,7 @@ int xe_mmio_wait32(struct xe_gt *gt, struct xe_reg reg, u32 val, u32 mask, u32 read; for (;;) { - read = xe_mmio_read32(gt, reg); + read = xe_mmio_read32_notrace(gt, reg); if ((read & mask) == val) { ret = 0; break; @@ -531,6 +601,8 @@ int xe_mmio_wait32(struct xe_gt *gt, struct xe_reg reg, u32 val, u32 mask, wait <<= 1; } + trace_xe_reg_rw(false, ®, read, sizeof(read)); + if (out_val) *out_val = read; diff --git a/drivers/gpu/drm/xe/xe_mmio.h b/drivers/gpu/drm/xe/xe_mmio.h index c06fcac2cc5b..02364ba9dda2 100644 --- a/drivers/gpu/drm/xe/xe_mmio.h +++ b/drivers/gpu/drm/xe/xe_mmio.h @@ -20,11 +20,16 @@ struct xe_device; #define GEN12_LMEM_BAR 2 int xe_mmio_init(struct xe_device *xe); +u8 xe_mmio_read8_notrace(struct xe_gt *gt, struct xe_reg reg); u8 xe_mmio_read8(struct xe_gt *gt, struct xe_reg reg); +void xe_mmio_write32_notrace(struct xe_gt *gt, struct xe_reg reg, u32 val); void xe_mmio_write32(struct xe_gt *gt, struct xe_reg reg, u32 val); +u32 xe_mmio_read32_notrace(struct xe_gt *gt, struct xe_reg reg); u32 xe_mmio_read32(struct xe_gt *gt, struct xe_reg reg); u32 xe_mmio_rmw32(struct xe_gt *gt, struct xe_reg reg, u32 clr, u32 set); +void xe_mmio_write64_notrace(struct xe_gt *gt, struct xe_reg reg, u64 val); void xe_mmio_write64(struct xe_gt *gt, struct xe_reg reg, u64 val); +u64 xe_mmio_read64_notrace(struct xe_gt *gt, struct xe_reg reg); u64 xe_mmio_read64(struct xe_gt *gt, struct xe_reg reg); int xe_mmio_write32_and_verify(struct xe_gt *gt, struct xe_reg reg, u32 val, u32 mask, u32 eval); diff --git a/drivers/gpu/drm/xe/xe_trace.h b/drivers/gpu/drm/xe/xe_trace.h index 2f8eb7ebe9a7..a5d514c190a2 100644 --- a/drivers/gpu/drm/xe/xe_trace.h +++ b/drivers/gpu/drm/xe/xe_trace.h @@ -12,6 +12,7 @@ #include #include +#include "regs/xe_reg_defs.h" #include "xe_bo_types.h" #include "xe_engine_types.h" #include "xe_gt_tlb_invalidation_types.h" @@ -507,6 +508,31 @@ DEFINE_EVENT(xe_vm, xe_vm_rebind_worker_exit, TP_ARGS(vm) ); +TRACE_EVENT(xe_reg_rw, + TP_PROTO(bool write, struct xe_reg *reg, u64 val, int len), + TP_ARGS(write, reg, val, len), + + TP_STRUCT__entry( + __field(u64, val) + __field(struct xe_reg *, reg) + __field(u16, write) + __field(u16, len) + ), + + TP_fast_assign( + __entry->val = (u64)val; + __entry->reg = reg; + __entry->write = write; + __entry->len = len; + ), + + TP_printk("%s reg=0x%x, len=%d, val=(0x%x, 0x%x)", + __entry->write ? "write" : "read", + __entry->reg->addr, __entry->len, + (u32)(__entry->val & 0xffffffff), + (u32)(__entry->val >> 32)) +); + TRACE_EVENT(xe_guc_ct_h2g_flow_control, TP_PROTO(u32 _head, u32 _tail, u32 size, u32 space, u32 len), TP_ARGS(_head, _tail, size, space, len), -- 2.34.1