[PATCH v5 6/6] drm/xe: Add reg read/write trace

[Gustavo Sousa]

Quoting Radhakrishna Sripada (2024-06-06 19:38:19-03:00)
>This will help debug register read/writes and provides
>a way to trace all the mmio transactions.
>
>v2: Fix kunit error
>v3: Print devid to help in multi-gpu setup
>v3: rebase and use variable sized variant to display
>    dev name(Gustavo)
>v4: Pass single argument to __asign_str to fix kunit error
>
>Cc: Gustavo Sousa <gustavo.sousa at intel.com>
>Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada at intel.com>
>---
> drivers/gpu/drm/xe/xe_mmio.c  | 24 ++++++++++++++++++++----
> drivers/gpu/drm/xe/xe_trace.h | 28 ++++++++++++++++++++++++++++
> 2 files changed, 48 insertions(+), 4 deletions(-)
>
>diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c
>index 7962eeb9adb7..feba35ad6a3e 100644
>--- a/drivers/gpu/drm/xe/xe_mmio.c
>+++ b/drivers/gpu/drm/xe/xe_mmio.c
>@@ -21,6 +21,8 @@
> #include "xe_gt_sriov_vf.h"
> #include "xe_macros.h"
> #include "xe_sriov.h"
>+#include "xe_tile.h"

The inclusion of xe_tile.h seems unrelated.

>+#include "xe_trace.h"
> 
> static void tiles_fini(void *arg)
> {
>@@ -124,16 +126,24 @@ u8 xe_mmio_read8(struct xe_gt *gt, struct xe_reg reg)
> {
>         struct xe_tile *tile = gt_to_tile(gt);
>         u32 addr = xe_mmio_adjusted_addr(gt, reg.addr);
>+        u8 val;
> 
>-        return readb((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+        val = readb((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+        trace_xe_reg_rw(gt, false, addr, val, sizeof(val));
>+
>+        return val;
> }
> 
> u16 xe_mmio_read16(struct xe_gt *gt, struct xe_reg reg)
> {
>         struct xe_tile *tile = gt_to_tile(gt);
>         u32 addr = xe_mmio_adjusted_addr(gt, reg.addr);
>+        u16 val;
>+
>+        val = readw((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+        trace_xe_reg_rw(gt, false, addr, val, sizeof(val));
> 
>-        return readw((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+        return val;
> }
> 
> void xe_mmio_write32(struct xe_gt *gt, struct xe_reg reg, u32 val)
>@@ -141,6 +151,7 @@ void xe_mmio_write32(struct xe_gt *gt, struct xe_reg reg, u32 val)
>         struct xe_tile *tile = gt_to_tile(gt);
>         u32 addr = xe_mmio_adjusted_addr(gt, reg.addr);
> 
>+        trace_xe_reg_rw(gt, true, addr, val, sizeof(val));
>         writel(val, (reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
> }
> 
>@@ -148,11 +159,16 @@ u32 xe_mmio_read32(struct xe_gt *gt, struct xe_reg reg)
> {
>         struct xe_tile *tile = gt_to_tile(gt);
>         u32 addr = xe_mmio_adjusted_addr(gt, reg.addr);
>+        u32 val;
> 
>         if (!reg.vf && IS_SRIOV_VF(gt_to_xe(gt)))
>-                return xe_gt_sriov_vf_read32(gt, reg);
>+                val = xe_gt_sriov_vf_read32(gt, reg);
>+        else
>+                val = readl((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+
>+        trace_xe_reg_rw(gt, false, addr, val, sizeof(val));
> 
>-        return readl((reg.ext ? tile->mmio_ext.regs : tile->mmio.regs) + addr);
>+        return val;
> }
> 
> u32 xe_mmio_rmw32(struct xe_gt *gt, struct xe_reg reg, u32 clr, u32 set)
>diff --git a/drivers/gpu/drm/xe/xe_trace.h b/drivers/gpu/drm/xe/xe_trace.h
>index f0b3477c0e1a..8d074a828d47 100644
>--- a/drivers/gpu/drm/xe/xe_trace.h
>+++ b/drivers/gpu/drm/xe/xe_trace.h
>@@ -347,6 +347,34 @@ DEFINE_EVENT(xe_hw_fence, xe_hw_fence_free,
>              TP_ARGS(fence)
> );
> 
>+TRACE_EVENT(xe_reg_rw,
>+        TP_PROTO(struct xe_gt *gt, bool write, u32 reg, u64 val, int len),
>+
>+        TP_ARGS(gt, write, reg, val, len),
>+
>+        TP_STRUCT__entry(
>+                __string(dev, __dev_name_gt(gt))
>+                __field(u64, val)
>+                __field(u32, reg)
>+                __field(u16, write)
>+                __field(u16, len)
>+                ),
>+
>+        TP_fast_assign(
>+                __assign_str(dev);
>+                __entry->val = (u64)val;

This cast seems unnecessary.

With those tweaks,

    Reviewed-by: Gustavo Sousa <gustavo.sousa at intel.com>

>+                __entry->reg = reg;
>+                __entry->write = write;
>+                __entry->len = len;
>+                ),
>+
>+        TP_printk("dev=%s, %s reg=0x%x, len=%d, val=(0x%x, 0x%x)",
>+                  __get_str(dev), __entry->write ? "write" : "read",
>+                  __entry->reg, __entry->len,
>+                  (u32)(__entry->val & 0xffffffff),
>+                  (u32)(__entry->val >> 32))
>+);
>+
> #endif
> 
> /* This part must be outside protection */
>-- 
>2.34.1
>