From mboxrd@z Thu Jan 1 00:00:00 1970 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 20A49F9D0FD for ; Wed, 15 Apr 2026 02:03:51 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 841B910E150; Wed, 15 Apr 2026 02:03:51 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="b0vX84Kr"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.13]) by gabe.freedesktop.org (Postfix) with ESMTPS id B0AD010E150 for ; Wed, 15 Apr 2026 02:03:50 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1776218630; x=1807754630; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=o9/klToUiaeOvmMWXFqjRg9oyuodCM0/mzSlDh+r7hc=; b=b0vX84KrBN8DmerD/SP5EDKi2fGphJXey816HIJRUftya5UfM1Zm7S7f n9+GPCcbyHqMcjD4rNGYXAbha59SBx3Hn92AVoJFj6+/D2ogxWk+YOoyW v3rWFzgSAelbwJZCznJrZICTUGngqj9zYho3KAu9xz79boyLm9Pzl3t9H okU8leeZP1txk/65ViXr9O+FeUHf3NfqaWchxEX6ajEP25yvojfwgoi/7 o9dqrzsKF9MmDZHBI55dqKuz0aizJRI+kxv6CmjQ4xZC32ye7dIGekf6W PnW+9mvqlGkgB+mSNwyK1ksjZMP0IOlZwJ5sWpGJuarHP2ZD9+UvYDuVj w==; X-CSE-ConnectionGUID: S5tkRx5ORtOODy1oPFEAyQ== X-CSE-MsgGUID: u4Kl1mHxRT2/TEekmgf/7A== X-IronPort-AV: E=McAfee;i="6800,10657,11759"; a="79773442" X-IronPort-AV: E=Sophos;i="6.23,179,1770624000"; d="scan'208";a="79773442" Received: from fmviesa005.fm.intel.com ([10.60.135.145]) by fmvoesa107.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 14 Apr 2026 19:03:50 -0700 X-CSE-ConnectionGUID: rwV9nxZKSHuUx92Wxl5p/w== X-CSE-MsgGUID: kyRXTQLSRvy21UuY3R8m7w== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.23,179,1770624000"; d="scan'208";a="235200876" Received: from orsosgc001.jf.intel.com ([10.88.27.185]) by fmviesa005-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 14 Apr 2026 19:03:49 -0700 From: Ashutosh Dixit To: intel-xe@lists.freedesktop.org Cc: Umesh Nerlige Ramappa Subject: [PATCH 1/3] drm/xe/oa: Use xe_map layer Date: Tue, 14 Apr 2026 19:03:40 -0700 Message-ID: <20260415020342.782987-2-ashutosh.dixit@intel.com> X-Mailer: git-send-email 2.48.1 In-Reply-To: <20260415020342.782987-1-ashutosh.dixit@intel.com> References: <20260415020342.782987-1-ashutosh.dixit@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit 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: , Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" OA code should have used xe_map layer to begin with. In CRI, the OA buffer can be located both in system and device memory. For these reasons, move OA code to use the xe_map layer when accessing the OA buffer. v2: Use xe_map layer and put_user in xe_oa_copy_to_user (Umesh) v3: To avoid performance impact in v2, revert to v1 but move xe_map_copy_to_user() to xe_map.h v4: Use bounce buffer and copy_to_user in xe_oa_copy_to_user Signed-off-by: Ashutosh Dixit --- drivers/gpu/drm/xe/xe_oa.c | 95 +++++++++++++++++++------------- drivers/gpu/drm/xe/xe_oa_types.h | 4 +- 2 files changed, 58 insertions(+), 41 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_oa.c b/drivers/gpu/drm/xe/xe_oa.c index 6337e671c97ae..4d13d8ef8103f 100644 --- a/drivers/gpu/drm/xe/xe_oa.c +++ b/drivers/gpu/drm/xe/xe_oa.c @@ -213,32 +213,40 @@ static u32 xe_oa_hw_tail_read(struct xe_oa_stream *stream) #define oa_report_header_64bit(__s) \ ((__s)->oa_buffer.format->header == HDR_64_BIT) -static u64 oa_report_id(struct xe_oa_stream *stream, void *report) +static u64 oa_report_id(struct xe_oa_stream *stream, u32 tail) { - return oa_report_header_64bit(stream) ? *(u64 *)report : *(u32 *)report; + struct iosys_map *map = &stream->oa_buffer.bo->vmap; + + return oa_report_header_64bit(stream) ? + xe_map_rd(stream->oa->xe, map, tail, u64) : + xe_map_rd(stream->oa->xe, map, tail, u32); } -static void oa_report_id_clear(struct xe_oa_stream *stream, u32 *report) +static void oa_report_id_clear(struct xe_oa_stream *stream, u32 head) { - if (oa_report_header_64bit(stream)) - *(u64 *)report = 0; - else - *report = 0; + struct iosys_map *map = &stream->oa_buffer.bo->vmap; + + oa_report_header_64bit(stream) ? + xe_map_wr(stream->oa->xe, map, head, u64, 0) : + xe_map_wr(stream->oa->xe, map, head, u32, 0); } -static u64 oa_timestamp(struct xe_oa_stream *stream, void *report) +static u64 oa_timestamp(struct xe_oa_stream *stream, u32 tail) { + struct iosys_map *map = &stream->oa_buffer.bo->vmap; + return oa_report_header_64bit(stream) ? - *((u64 *)report + 1) : - *((u32 *)report + 1); + xe_map_rd(stream->oa->xe, map, tail + 2, u64) : + xe_map_rd(stream->oa->xe, map, tail + 1, u32); } -static void oa_timestamp_clear(struct xe_oa_stream *stream, u32 *report) +static void oa_timestamp_clear(struct xe_oa_stream *stream, u32 head) { - if (oa_report_header_64bit(stream)) - *(u64 *)&report[2] = 0; - else - report[1] = 0; + struct iosys_map *map = &stream->oa_buffer.bo->vmap; + + oa_report_header_64bit(stream) ? + xe_map_wr(stream->oa->xe, map, head + 2, u64, 0) : + xe_map_wr(stream->oa->xe, map, head + 1, u32, 0); } static bool xe_oa_buffer_check_unlocked(struct xe_oa_stream *stream) @@ -275,9 +283,7 @@ static bool xe_oa_buffer_check_unlocked(struct xe_oa_stream *stream) * they were written. If not : (╯°□°）╯︵ ┻━┻ */ while (xe_oa_circ_diff(stream, tail, stream->oa_buffer.tail) >= report_size) { - void *report = stream->oa_buffer.vaddr + tail; - - if (oa_report_id(stream, report) || oa_timestamp(stream, report)) + if (oa_report_id(stream, tail) || oa_timestamp(stream, tail)) break; tail = xe_oa_circ_diff(stream, tail, report_size); @@ -311,30 +317,35 @@ static enum hrtimer_restart xe_oa_poll_check_timer_cb(struct hrtimer *hrtimer) return HRTIMER_RESTART; } +static unsigned long +xe_oa_copy_to_user(struct xe_oa_stream *stream, void __user *dst, u32 head, u32 len) +{ + xe_map_memcpy_from(stream->oa->xe, stream->oa_buffer.bounce, + &stream->oa_buffer.bo->vmap, head, len); + return copy_to_user(dst, stream->oa_buffer.bounce, len); +} + static int xe_oa_append_report(struct xe_oa_stream *stream, char __user *buf, - size_t count, size_t *offset, const u8 *report) + size_t count, size_t *offset, u32 head) { int report_size = stream->oa_buffer.format->size; int report_size_partial; - u8 *oa_buf_end; if ((count - *offset) < report_size) return -ENOSPC; buf += *offset; - oa_buf_end = stream->oa_buffer.vaddr + stream->oa_buffer.circ_size; - report_size_partial = oa_buf_end - report; + report_size_partial = stream->oa_buffer.circ_size - head; if (report_size_partial < report_size) { - if (copy_to_user(buf, report, report_size_partial)) + if (xe_oa_copy_to_user(stream, buf, head, report_size_partial)) return -EFAULT; buf += report_size_partial; - if (copy_to_user(buf, stream->oa_buffer.vaddr, - report_size - report_size_partial)) + if (xe_oa_copy_to_user(stream, buf, 0, report_size - report_size_partial)) return -EFAULT; - } else if (copy_to_user(buf, report, report_size)) { + } else if (xe_oa_copy_to_user(stream, buf, head, report_size)) { return -EFAULT; } @@ -347,7 +358,6 @@ static int xe_oa_append_reports(struct xe_oa_stream *stream, char __user *buf, size_t count, size_t *offset) { int report_size = stream->oa_buffer.format->size; - u8 *oa_buf_base = stream->oa_buffer.vaddr; u32 gtt_offset = xe_bo_ggtt_addr(stream->oa_buffer.bo); size_t start_offset = *offset; unsigned long flags; @@ -364,26 +374,24 @@ static int xe_oa_append_reports(struct xe_oa_stream *stream, char __user *buf, for (; xe_oa_circ_diff(stream, tail, head); head = xe_oa_circ_incr(stream, head, report_size)) { - u8 *report = oa_buf_base + head; - - ret = xe_oa_append_report(stream, buf, count, offset, report); + ret = xe_oa_append_report(stream, buf, count, offset, head); if (ret) break; if (!(stream->oa_buffer.circ_size % report_size)) { /* Clear out report id and timestamp to detect unlanded reports */ - oa_report_id_clear(stream, (void *)report); - oa_timestamp_clear(stream, (void *)report); + oa_report_id_clear(stream, head); + oa_timestamp_clear(stream, head); } else { - u8 *oa_buf_end = stream->oa_buffer.vaddr + stream->oa_buffer.circ_size; - u32 part = oa_buf_end - report; + struct iosys_map *map = &stream->oa_buffer.bo->vmap; + u32 part = stream->oa_buffer.circ_size - head; /* Zero out the entire report */ if (report_size <= part) { - memset(report, 0, report_size); + xe_map_memset(stream->oa->xe, map, head, 0, report_size); } else { - memset(report, 0, part); - memset(oa_buf_base, 0, report_size - part); + xe_map_memset(stream->oa->xe, map, head, 0, part); + xe_map_memset(stream->oa->xe, map, 0, 0, report_size - part); } } } @@ -436,7 +444,8 @@ static void xe_oa_init_oa_buffer(struct xe_oa_stream *stream) spin_unlock_irqrestore(&stream->oa_buffer.ptr_lock, flags); /* Zero out the OA buffer since we rely on zero report id and timestamp fields */ - memset(stream->oa_buffer.vaddr, 0, xe_bo_size(stream->oa_buffer.bo)); + xe_map_memset(stream->oa->xe, &stream->oa_buffer.bo->vmap, 0, 0, + xe_bo_size(stream->oa_buffer.bo)); } static u32 __format_to_oactrl(const struct xe_oa_format *format, int counter_sel_mask) @@ -699,6 +708,7 @@ static int num_lri_dwords(int num_regs) static void xe_oa_free_oa_buffer(struct xe_oa_stream *stream) { xe_bo_unpin_map_no_vm(stream->oa_buffer.bo); + kfree(stream->oa_buffer.bounce); } static void xe_oa_free_configs(struct xe_oa_stream *stream) @@ -889,9 +899,16 @@ static int xe_oa_alloc_oa_buffer(struct xe_oa_stream *stream, size_t size) return PTR_ERR(bo); stream->oa_buffer.bo = bo; + /* mmap implementation requires OA buffer to be in system memory */ xe_assert(stream->oa->xe, bo->vmap.is_iomem == 0); - stream->oa_buffer.vaddr = bo->vmap.vaddr; + + stream->oa_buffer.bounce = kmalloc(stream->oa_buffer.format->size, GFP_KERNEL); + if (!stream->oa_buffer.bounce) { + xe_bo_unpin_map_no_vm(stream->oa_buffer.bo); + return -ENOMEM; + } + return 0; } diff --git a/drivers/gpu/drm/xe/xe_oa_types.h b/drivers/gpu/drm/xe/xe_oa_types.h index b03ffd5134834..99a04b8d586b5 100644 --- a/drivers/gpu/drm/xe/xe_oa_types.h +++ b/drivers/gpu/drm/xe/xe_oa_types.h @@ -162,8 +162,8 @@ struct xe_oa_buffer { /** @format: xe_bo backing the OA buffer */ struct xe_bo *bo; - /** @vaddr: mapped vaddr of the OA buffer */ - u8 *vaddr; + /** @bounce: bounce buffer used with xe_map layer */ + void *bounce; /** @ptr_lock: Lock protecting reads/writes to head/tail pointers */ spinlock_t ptr_lock; -- 2.48.1