Skip to contentSkip to conversation

More sm501 fixes and optimisations

PATCH In review BALATON Zoltan 3 participants 15 messages Jun 20, 2020 – Jun 30, 2020

lore
Share & tools
lore
#0 lore raw

v3 cover More sm501 fixes and optimisations

Version 3: added R-b tags and new patch to fix overlap checks. Hopefully I got it right this time, unless there are any more comments this should be good to go.

BALATON Zoltan (9):
sm501: Fix bounds checks
sm501: Drop unneded variable
sm501: Ignore no-op blits
sm501: Introduce variable for commonly used value for better
readability
sm501: Optimise 1 pixel 2d ops
sm501: Use stn_he_p/ldn_he_p instead of switch/case
sm501: Do not allow guest to set invalid format
sm501: Convert debug printfs to traces
sm501: Fix and optimize overlap check

 hw/display/sm501.c      | 157 +++++++++++++++++++---------------------
 hw/display/trace-events |  12 +++
 2 files changed, 87 insertions(+), 82 deletions(-)
Signature · 1 line
2.21.3
#1 lore raw

PATCH v3 1/9 sm501: Fix bounds checks

We don't need to add width to pitch when calculating last point, that would reject valid ops within the card's local_mem.

Fixes b15a22bbcbe6a78dc3d88fe3134985e4cdd87de4
---
 hw/display/sm501.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
hw/display/sm501.c +4 -4
index a7fc08c52b..5ceee4166f 100644
@@ -723,8 +723,8 @@ static void sm501_2d_operation(SM501State *s)
         dst_y -= height - 1;
     }
 
-    if (dst_base >= get_local_mem_size(s) || dst_base +
-        (dst_x + width + (dst_y + height) * (dst_pitch + width)) *
+    if (dst_base >= get_local_mem_size(s) ||
+        dst_base + (dst_x + width + (dst_y + height) * dst_pitch) *
         (1 << format) >= get_local_mem_size(s)) {
         qemu_log_mask(LOG_GUEST_ERROR, "sm501: 2D op dest is outside vram.\n");
         return;
@@ -749,8 +749,8 @@ static void sm501_2d_operation(SM501State *s)
             src_y -= height - 1;
         }
 
-        if (src_base >= get_local_mem_size(s) || src_base +
-            (src_x + width + (src_y + height) * (src_pitch + width)) *
+        if (src_base >= get_local_mem_size(s) ||
+            src_base + (src_x + width + (src_y + height) * src_pitch) *
             (1 << format) >= get_local_mem_size(s)) {
             qemu_log_mask(LOG_GUEST_ERROR,
                           "sm501: 2D op src is outside vram.\n");
Signature · 1 line
2.21.3
#2 lore raw

PATCH v3 2/9 sm501: Drop unneded variable

In reply to BALATON Zoltan

We don't need a separate variable to keep track if we allocated memory that needs to be freed as we can test the pointer itself.

---
 hw/display/sm501.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
hw/display/sm501.c +2 -3
index 5ceee4166f..5d2e019575 100644
@@ -796,13 +796,12 @@ static void sm501_2d_operation(SM501State *s)
             de = db + width + height * (width + dst_pitch);
             if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
                 /* regions may overlap: copy via temporary */
-                int free_buf = 0, llb = width * (1 << format);
+                int llb = width * (1 << format);
                 int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
                 if (tmp_stride * sizeof(uint32_t) * height > sizeof(tmp_buf)) {
                     tmp = g_malloc(tmp_stride * sizeof(uint32_t) * height);
-                    free_buf = 1;
                 }
                 pixman_blt((uint32_t *)&s->local_mem[src_base], tmp,
                            src_pitch * (1 << format) / sizeof(uint32_t),
@@ -813,7 +812,7 @@ static void sm501_2d_operation(SM501State *s)
                            dst_pitch * (1 << format) / sizeof(uint32_t),
                            8 * (1 << format), 8 * (1 << format),
                            0, 0, dst_x, dst_y, width, height);
-                if (free_buf) {
+                if (tmp != tmp_buf) {
                     g_free(tmp);
                 }
             } else {
Signature · 1 line
2.21.3
#3 lore raw

PATCH v3 3/9 sm501: Ignore no-op blits

In reply to BALATON Zoltan

Some guests seem to try source copy blits with same source and dest which are no-op so avoid calling pixman for these.

---
 hw/display/sm501.c | 5 +++++
 1 file changed, 5 insertions(+)
hw/display/sm501.c +5 -0
index 5d2e019575..ad5a62bfab 100644
@@ -788,6 +788,11 @@ static void sm501_2d_operation(SM501State *s)
                               (rop2_source_is_pattern ?
                                   " with pattern source" : ""));
             }
+            /* Ignore no-op blits, some guests seem to do this */
+            if (src_base == dst_base && src_pitch == dst_pitch &&
+                src_x == dst_x && src_y == dst_y) {
+                break;
+            }
             /* Check for overlaps, this could be made more exact */
             uint32_t sb, se, db, de;
             sb = src_base + src_x + src_y * (width + src_pitch);
Signature · 1 line
2.21.3
#4 lore raw

PATCH v3 4/9 sm501: Introduce variable for commonly used value for better readability

In reply to BALATON Zoltan

The bytes per pixel value can be calculated from format but it's used freqently enough (and will be used more in subseqent patches) so store it in a variable for better readabilty. Also drop some unneded 0x prefix around where new variable is defined.

---
 hw/display/sm501.c | 41 +++++++++++++++++++++--------------------
 1 file changed, 21 insertions(+), 20 deletions(-)
hw/display/sm501.c +21 -20
index ad5a62bfab..3ced2c42db 100644
@@ -684,10 +684,11 @@ static void sm501_2d_operation(SM501State *s)
 {
     int cmd = (s->twoD_control >> 16) & 0x1F;
     int rtl = s->twoD_control & BIT(27);
-    int format = (s->twoD_stretch >> 20) & 0x3;
-    int rop_mode = (s->twoD_control >> 15) & 0x1; /* 1 for rop2, else rop3 */
+    int format = (s->twoD_stretch >> 20) & 3;
+    int bypp = 1 << format; /* bytes per pixel */
+    int rop_mode = (s->twoD_control >> 15) & 1; /* 1 for rop2, else rop3 */
     /* 1 if rop2 source is the pattern, otherwise the source is the bitmap */
-    int rop2_source_is_pattern = (s->twoD_control >> 14) & 0x1;
+    int rop2_source_is_pattern = (s->twoD_control >> 14) & 1;
     int rop = s->twoD_control & 0xFF;
     unsigned int dst_x = (s->twoD_destination >> 16) & 0x01FFF;
     unsigned int dst_y = s->twoD_destination & 0xFFFF;
@@ -724,8 +725,8 @@ static void sm501_2d_operation(SM501State *s)
     }
 
     if (dst_base >= get_local_mem_size(s) ||
-        dst_base + (dst_x + width + (dst_y + height) * dst_pitch) *
-        (1 << format) >= get_local_mem_size(s)) {
+        dst_base + (dst_x + width + (dst_y + height) * dst_pitch) * bypp >=
+        get_local_mem_size(s)) {
         qemu_log_mask(LOG_GUEST_ERROR, "sm501: 2D op dest is outside vram.\n");
         return;
     }
@@ -750,8 +751,8 @@ static void sm501_2d_operation(SM501State *s)
         }
 
         if (src_base >= get_local_mem_size(s) ||
-            src_base + (src_x + width + (src_y + height) * src_pitch) *
-            (1 << format) >= get_local_mem_size(s)) {
+            src_base + (src_x + width + (src_y + height) * src_pitch) * bypp >=
+            get_local_mem_size(s)) {
             qemu_log_mask(LOG_GUEST_ERROR,
                           "sm501: 2D op src is outside vram.\n");
             return;
@@ -763,8 +764,8 @@ static void sm501_2d_operation(SM501State *s)
             uint8_t *d = s->local_mem + dst_base;
 
             for (y = 0; y < height; y++) {
-                i = (dst_x + (dst_y + y) * dst_pitch) * (1 << format);
-                for (x = 0; x < width; x++, i += (1 << format)) {
+                i = (dst_x + (dst_y + y) * dst_pitch) * bypp;
+                for (x = 0; x < width; x++, i += bypp) {
                     switch (format) {
                     case 0:
                         d[i] = ~d[i];
@@ -801,7 +802,7 @@ static void sm501_2d_operation(SM501State *s)
             de = db + width + height * (width + dst_pitch);
             if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
                 /* regions may overlap: copy via temporary */
-                int llb = width * (1 << format);
+                int llb = width * bypp;
                 int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
@@ -809,13 +810,13 @@ static void sm501_2d_operation(SM501State *s)
                     tmp = g_malloc(tmp_stride * sizeof(uint32_t) * height);
                 }
                 pixman_blt((uint32_t *)&s->local_mem[src_base], tmp,
-                           src_pitch * (1 << format) / sizeof(uint32_t),
-                           tmp_stride, 8 * (1 << format), 8 * (1 << format),
+                           src_pitch * bypp / sizeof(uint32_t),
+                           tmp_stride, 8 * bypp, 8 * bypp,
                            src_x, src_y, 0, 0, width, height);
                 pixman_blt(tmp, (uint32_t *)&s->local_mem[dst_base],
                            tmp_stride,
-                           dst_pitch * (1 << format) / sizeof(uint32_t),
-                           8 * (1 << format), 8 * (1 << format),
+                           dst_pitch * bypp / sizeof(uint32_t),
+                           8 * bypp, 8 * bypp,
                            0, 0, dst_x, dst_y, width, height);
                 if (tmp != tmp_buf) {
                     g_free(tmp);
@@ -823,9 +824,9 @@ static void sm501_2d_operation(SM501State *s)
             } else {
                 pixman_blt((uint32_t *)&s->local_mem[src_base],
                            (uint32_t *)&s->local_mem[dst_base],
-                           src_pitch * (1 << format) / sizeof(uint32_t),
-                           dst_pitch * (1 << format) / sizeof(uint32_t),
-                           8 * (1 << format), 8 * (1 << format),
+                           src_pitch * bypp / sizeof(uint32_t),
+                           dst_pitch * bypp / sizeof(uint32_t),
+                           8 * bypp, 8 * bypp,
                            src_x, src_y, dst_x, dst_y, width, height);
             }
         }
@@ -842,8 +843,8 @@ static void sm501_2d_operation(SM501State *s)
         }
 
         pixman_fill((uint32_t *)&s->local_mem[dst_base],
-                    dst_pitch * (1 << format) / sizeof(uint32_t),
-                    8 * (1 << format), dst_x, dst_y, width, height, color);
+                    dst_pitch * bypp / sizeof(uint32_t),
+                    8 * bypp, dst_x, dst_y, width, height, color);
         break;
     }
     default:
@@ -855,7 +856,7 @@ static void sm501_2d_operation(SM501State *s)
     if (dst_base >= get_fb_addr(s, crt) &&
         dst_base <= get_fb_addr(s, crt) + fb_len) {
         int dst_len = MIN(fb_len, ((dst_y + height - 1) * dst_pitch +
-                          dst_x + width) * (1 << format));
+                          dst_x + width) * bypp);
         if (dst_len) {
             memory_region_set_dirty(&s->local_mem_region, dst_base, dst_len);
         }
Signature · 1 line
2.21.3
#5 lore raw

PATCH v3 5/9 sm501: Optimise 1 pixel 2d ops

In reply to BALATON Zoltan

Some guests do 1x1 blits which is faster to do directly than calling a function for it so avoid overhead in this case.

---
 hw/display/sm501.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
hw/display/sm501.c +16 -3
index 3ced2c42db..2098e69810 100644
@@ -794,6 +794,14 @@ static void sm501_2d_operation(SM501State *s)
                 src_x == dst_x && src_y == dst_y) {
                 break;
             }
+            /* Some clients also do 1 pixel blits, avoid overhead for these */
+            if (width == 1 && height == 1) {
+                unsigned int si = (src_x + src_y * src_pitch) * bypp;
+                unsigned int di = (dst_x + dst_y * dst_pitch) * bypp;
+                stn_he_p(&s->local_mem[dst_base + di], bypp,
+                         ldn_he_p(&s->local_mem[src_base + si], bypp));
+                break;
+            }
             /* Check for overlaps, this could be made more exact */
             uint32_t sb, se, db, de;
             sb = src_base + src_x + src_y * (width + src_pitch);
@@ -842,9 +850,14 @@ static void sm501_2d_operation(SM501State *s)
             color = cpu_to_le16(color);
         }
 
-        pixman_fill((uint32_t *)&s->local_mem[dst_base],
-                    dst_pitch * bypp / sizeof(uint32_t),
-                    8 * bypp, dst_x, dst_y, width, height, color);
+        if (width == 1 && height == 1) {
+            unsigned int i = (dst_x + dst_y * dst_pitch) * bypp;
+            stn_he_p(&s->local_mem[dst_base + i], bypp, color);
+        } else {
+            pixman_fill((uint32_t *)&s->local_mem[dst_base],
+                        dst_pitch * bypp / sizeof(uint32_t),
+                        8 * bypp, dst_x, dst_y, width, height, color);
+        }
         break;
     }
     default:
Signature · 1 line
2.21.3
#6 lore raw

PATCH v3 6/9 sm501: Use stn_he_p/ldn_he_p instead of switch/case

In reply to BALATON Zoltan

Instead of open coding op with different sizes using a switch and type casting it can be written more compactly using stn_he_p/ldn_he_p.

---
 hw/display/sm501.c | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)
hw/display/sm501.c +1 -11
index 2098e69810..6349f31e64 100644
@@ -766,17 +766,7 @@ static void sm501_2d_operation(SM501State *s)
             for (y = 0; y < height; y++) {
                 i = (dst_x + (dst_y + y) * dst_pitch) * bypp;
                 for (x = 0; x < width; x++, i += bypp) {
-                    switch (format) {
-                    case 0:
-                        d[i] = ~d[i];
-                        break;
-                    case 1:
-                        *(uint16_t *)&d[i] = ~*(uint16_t *)&d[i];
-                        break;
-                    case 2:
-                        *(uint32_t *)&d[i] = ~*(uint32_t *)&d[i];
-                        break;
-                    }
+                    stn_he_p(&d[i], bypp, ~ldn_he_p(&d[i], bypp));
                 }
             }
         } else {
Signature · 1 line
2.21.3
#7 lore raw

PATCH v3 7/9 sm501: Do not allow guest to set invalid format

In reply to BALATON Zoltan

Prevent guest setting invalid format value that might trip checks in sm501_2d_operation().

---
 hw/display/sm501.c | 3 +++
 1 file changed, 3 insertions(+)
hw/display/sm501.c +3 -0
index 6349f31e64..7e4c042d52 100644
@@ -1503,6 +1503,9 @@ static void sm501_2d_engine_write(void *opaque, hwaddr addr,
         s->twoD_background = value;
         break;
     case SM501_2D_STRETCH:
+        if (((value >> 20) & 3) == 3) {
+            value &= ~BIT(20);
+        }
         s->twoD_stretch = value;
         break;
     case SM501_2D_COLOR_COMPARE:
Signature · 1 line
2.21.3
#8 lore raw

PATCH v3 8/9 sm501: Convert debug printfs to traces

In reply to BALATON Zoltan
---
 hw/display/sm501.c      | 50 +++++++++++------------------------------
 hw/display/trace-events | 12 ++++++++++
 2 files changed, 25 insertions(+), 37 deletions(-)
hw/display/sm501.c +13 -37
index 7e4c042d52..2db347dcbc 100644
@@ -39,15 +39,7 @@
 #include "qemu/range.h"
 #include "ui/pixel_ops.h"
 #include "qemu/bswap.h"
-
-/*#define DEBUG_SM501*/
-/*#define DEBUG_BITBLT*/
-
-#ifdef DEBUG_SM501
-#define SM501_DPRINTF(fmt, ...) printf(fmt, ## __VA_ARGS__)
-#else
-#define SM501_DPRINTF(fmt, ...) do {} while (0)
-#endif
+#include "trace.h"
 
 #define MMIO_BASE_OFFSET 0x3e00000
 #define MMIO_SIZE 0x200000
@@ -871,7 +863,6 @@ static uint64_t sm501_system_config_read(void *opaque, hwaddr addr,
 {
     SM501State *s = (SM501State *)opaque;
     uint32_t ret = 0;
-    SM501_DPRINTF("sm501 system config regs : read addr=%x\n", (int)addr);
 
     switch (addr) {
     case SM501_SYSTEM_CONTROL:
@@ -923,7 +914,7 @@ static uint64_t sm501_system_config_read(void *opaque, hwaddr addr,
         qemu_log_mask(LOG_UNIMP, "sm501: not implemented system config"
                       "register read. addr=%" HWADDR_PRIx "\n", addr);
     }
-
+    trace_sm501_system_config_read(addr, ret);
     return ret;
 }
 
@@ -931,9 +922,8 @@ static void sm501_system_config_write(void *opaque, hwaddr addr,
                                       uint64_t value, unsigned size)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 system config regs : write addr=%x, val=%x\n",
-                  (uint32_t)addr, (uint32_t)value);
 
+    trace_sm501_system_config_write((uint32_t)addr, (uint32_t)value);
     switch (addr) {
     case SM501_SYSTEM_CONTROL:
         s->system_control &= 0x10DB0000;
@@ -1019,9 +1009,7 @@ static uint64_t sm501_i2c_read(void *opaque, hwaddr addr, unsigned size)
         qemu_log_mask(LOG_UNIMP, "sm501 i2c : not implemented register read."
                       " addr=0x%" HWADDR_PRIx "\n", addr);
     }
-
-    SM501_DPRINTF("sm501 i2c regs : read addr=%" HWADDR_PRIx " val=%x\n",
-                  addr, ret);
+    trace_sm501_i2c_read((uint32_t)addr, ret);
     return ret;
 }
 
@@ -1029,9 +1017,8 @@ static void sm501_i2c_write(void *opaque, hwaddr addr, uint64_t value,
                             unsigned size)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 i2c regs : write addr=%" HWADDR_PRIx
-                  " val=%" PRIx64 "\n", addr, value);
 
+    trace_sm501_i2c_write((uint32_t)addr, (uint32_t)value);
     switch (addr) {
     case SM501_I2C_BYTE_COUNT:
         s->i2c_byte_count = value & 0xf;
@@ -1045,25 +1032,19 @@ static void sm501_i2c_write(void *opaque, hwaddr addr, uint64_t value,
                 s->i2c_status |= (res ? SM501_I2C_STATUS_ERROR : 0);
                 if (!res) {
                     int i;
-                    SM501_DPRINTF("sm501 i2c : transferring %d bytes to 0x%x\n",
-                                  s->i2c_byte_count + 1, s->i2c_addr >> 1);
                     for (i = 0; i <= s->i2c_byte_count; i++) {
                         res = i2c_send_recv(s->i2c_bus, &s->i2c_data[i],
                                             !(s->i2c_addr & 1));
                         if (res) {
-                            SM501_DPRINTF("sm501 i2c : transfer failed"
-                                          " i=%d, res=%d\n", i, res);
                             s->i2c_status |= SM501_I2C_STATUS_ERROR;
                             return;
                         }
                     }
                     if (i) {
-                        SM501_DPRINTF("sm501 i2c : transferred %d bytes\n", i);
                         s->i2c_status = SM501_I2C_STATUS_COMPLETE;
                     }
                 }
             } else {
-                SM501_DPRINTF("sm501 i2c : end transfer\n");
                 i2c_end_transfer(s->i2c_bus);
                 s->i2c_status &= ~SM501_I2C_STATUS_ERROR;
             }
@@ -1103,7 +1084,8 @@ static const MemoryRegionOps sm501_i2c_ops = {
 static uint32_t sm501_palette_read(void *opaque, hwaddr addr)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 palette read addr=%x\n", (int)addr);
+
+    trace_sm501_palette_read((uint32_t)addr);
 
     /* TODO : consider BYTE/WORD access */
     /* TODO : consider endian */
@@ -1116,8 +1098,8 @@ static void sm501_palette_write(void *opaque, hwaddr addr,
                                 uint32_t value)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 palette write addr=%x, val=%x\n",
-                  (int)addr, value);
+
+    trace_sm501_palette_write((uint32_t)addr, value);
 
     /* TODO : consider BYTE/WORD access */
     /* TODO : consider endian */
@@ -1132,7 +1114,6 @@ static uint64_t sm501_disp_ctrl_read(void *opaque, hwaddr addr,
 {
     SM501State *s = (SM501State *)opaque;
     uint32_t ret = 0;
-    SM501_DPRINTF("sm501 disp ctrl regs : read addr=%x\n", (int)addr);
 
     switch (addr) {
 
@@ -1237,7 +1218,7 @@ static uint64_t sm501_disp_ctrl_read(void *opaque, hwaddr addr,
         qemu_log_mask(LOG_UNIMP, "sm501: not implemented disp ctrl register "
                       "read. addr=%" HWADDR_PRIx "\n", addr);
     }
-
+    trace_sm501_disp_ctrl_read((uint32_t)addr, ret);
     return ret;
 }
 
@@ -1245,9 +1226,8 @@ static void sm501_disp_ctrl_write(void *opaque, hwaddr addr,
                                   uint64_t value, unsigned size)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 disp ctrl regs : write addr=%x, val=%x\n",
-                  (unsigned)addr, (unsigned)value);
 
+    trace_sm501_disp_ctrl_write((uint32_t)addr, (uint32_t)value);
     switch (addr) {
     case SM501_DC_PANEL_CONTROL:
         s->dc_panel_control = value & 0x0FFF73FF;
@@ -1392,7 +1372,6 @@ static uint64_t sm501_2d_engine_read(void *opaque, hwaddr addr,
 {
     SM501State *s = (SM501State *)opaque;
     uint32_t ret = 0;
-    SM501_DPRINTF("sm501 2d engine regs : read addr=%x\n", (int)addr);
 
     switch (addr) {
     case SM501_2D_SOURCE:
@@ -1462,7 +1441,7 @@ static uint64_t sm501_2d_engine_read(void *opaque, hwaddr addr,
         qemu_log_mask(LOG_UNIMP, "sm501: not implemented disp ctrl register "
                       "read. addr=%" HWADDR_PRIx "\n", addr);
     }
-
+    trace_sm501_2d_engine_read((uint32_t)addr, ret);
     return ret;
 }
 
@@ -1470,9 +1449,8 @@ static void sm501_2d_engine_write(void *opaque, hwaddr addr,
                                   uint64_t value, unsigned size)
 {
     SM501State *s = (SM501State *)opaque;
-    SM501_DPRINTF("sm501 2d engine regs : write addr=%x, val=%x\n",
-                  (unsigned)addr, (unsigned)value);
 
+    trace_sm501_2d_engine_write((uint32_t)addr, (uint32_t)value);
     switch (addr) {
     case SM501_2D_SOURCE:
         s->twoD_source = value;
@@ -1830,8 +1808,6 @@ static void sm501_init(SM501State *s, DeviceState *dev,
                        uint32_t local_mem_bytes)
 {
     s->local_mem_size_index = get_local_mem_size_index(local_mem_bytes);
-    SM501_DPRINTF("sm501 local mem size=%x. index=%d\n", get_local_mem_size(s),
-                  s->local_mem_size_index);
 
     /* local memory */
     memory_region_init_ram(&s->local_mem_region, OBJECT(dev), "sm501.local",
hw/display/trace-events +12 -0
index 72d4c9812c..970d6bac5d 100644
@@ -161,3 +161,15 @@ cg3_write(uint32_t addr, uint32_t val, unsigned size) "write addr:0x%06"PRIx32"
 # dpcd.c
 dpcd_read(uint32_t addr, uint8_t val) "read addr:0x%"PRIx32" val:0x%02x"
 dpcd_write(uint32_t addr, uint8_t val) "write addr:0x%"PRIx32" val:0x%02x"
+
+# sm501.c
+sm501_system_config_read(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_system_config_write(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_i2c_read(uint32_t addr, uint8_t val) "addr=0x%x, val=0x%x"
+sm501_i2c_write(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_palette_read(uint32_t addr) "addr=0x%x"
+sm501_palette_write(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_disp_ctrl_read(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_disp_ctrl_write(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_2d_engine_read(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
+sm501_2d_engine_write(uint32_t addr, uint32_t val) "addr=0x%x, val=0x%x"
Signature · 1 line
2.21.3
#9 lore raw

PATCH v3 9/9 sm501: Fix and optimize overlap check

In reply to BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

---
 hw/display/sm501.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
hw/display/sm501.c +16 -10
index 2db347dcbc..e7c69bf7fd 100644
@@ -690,6 +690,7 @@ static void sm501_2d_operation(SM501State *s)
     unsigned int dst_pitch = (s->twoD_pitch >> 16) & 0x1FFF;
     int crt = (s->dc_crt_control & SM501_DC_CRT_CONTROL_SEL) ? 1 : 0;
     int fb_len = get_width(s, crt) * get_height(s, crt) * get_bpp(s, crt);
+    bool overlap = false;
 
     if ((s->twoD_stretch >> 16) & 0xF) {
         qemu_log_mask(LOG_UNIMP, "sm501: only XY addressing is supported.\n");
@@ -784,16 +785,21 @@ static void sm501_2d_operation(SM501State *s)
                          ldn_he_p(&s->local_mem[src_base + si], bypp));
                 break;
             }
-            /* Check for overlaps, this could be made more exact */
-            uint32_t sb, se, db, de;
-            sb = src_base + src_x + src_y * (width + src_pitch);
-            se = sb + width + height * (width + src_pitch);
-            db = dst_base + dst_x + dst_y * (width + dst_pitch);
-            de = db + width + height * (width + dst_pitch);
-            if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
-                /* regions may overlap: copy via temporary */
-                int llb = width * bypp;
-                int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
+            /* If reverse blit do simple check for overlaps */
+            if (rtl && src_base == dst_base && src_pitch == dst_pitch) {
+                overlap = (src_x < dst_x + width && src_x + width > dst_x &&
+                           src_y < dst_y + height && src_y + height > dst_y);
+            } else if (rtl) {
+                unsigned int sb, se, db, de;
+                sb = src_base + (src_x + src_y * src_pitch) * bypp;
+                se = sb + (width + height * src_pitch) * bypp;
+                db = dst_base + (dst_x + dst_y * dst_pitch) * bypp;
+                de = db + (width + height * dst_pitch) * bypp;
+                overlap = (db >= sb && db <= se) || (de >= sb && de <= se);
+            }
+            if (overlap) {
+                /* pixman can't do reverse blit: copy via temporary */
+                int tmp_stride = DIV_ROUND_UP(width * bypp, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
                 if (tmp_stride * sizeof(uint32_t) * height > sizeof(tmp_buf)) {
Signature · 1 line
2.21.3
#10 lore raw

Re: [PATCH v3 9/9] sm501: Fix and optimize overlap check

Quoted 32 lines · BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

---
 hw/display/sm501.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
hw/display/sm501.c +1 -4
index 2db347dcbc..e7c69bf7fd 100644
@@ -690,6 +690,7 @@ static void sm501_2d_operation(SM501State *s)
     unsigned int dst_pitch = (s->twoD_pitch >> 16) & 0x1FFF;
     int crt = (s->dc_crt_control & SM501_DC_CRT_CONTROL_SEL) ? 1 : 0;
     int fb_len = get_width(s, crt) * get_height(s, crt) * get_bpp(s, crt);
+    bool overlap = false;
 
     if ((s->twoD_stretch >> 16) & 0xF) {
         qemu_log_mask(LOG_UNIMP, "sm501: only XY addressing is supported.\n");
@@ -784,16 +785,21 @@ static void sm501_2d_operation(SM501State *s)
                          ldn_he_p(&s->local_mem[src_base + si], bypp));
                 break;
             }
-            /* Check for overlaps, this could be made more exact */
-            uint32_t sb, se, db, de;
-            sb = src_base + src_x + src_y * (width + src_pitch);
-            se = sb + width + height * (width + src_pitch);

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 791

Quoted 10 lines
hw/display/sm501.c +4 -6
-            db = dst_base + dst_x + dst_y * (width + dst_pitch);
-            de = db + width + height * (width + dst_pitch);
-            if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
-                /* regions may overlap: copy via temporary */
-                int llb = width * bypp;
-                int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
+            /* If reverse blit do simple check for overlaps */
+            if (rtl && src_base == dst_base && src_pitch == dst_pitch) {
+                overlap = (src_x < dst_x + width && src_x + width > dst_x &&
+                           src_y < dst_y + height && src_y + height > dst_y);

This part looks good...

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 798

Quoted 7 lines
hw/display/sm501.c +7 -0
+            } else if (rtl) {
+                unsigned int sb, se, db, de;
+                sb = src_base + (src_x + src_y * src_pitch) * bypp;
+                se = sb + (width + height * src_pitch) * bypp;
+                db = dst_base + (dst_x + dst_y * dst_pitch) * bypp;
+                de = db + (width + height * dst_pitch) * bypp;
+                overlap = (db >= sb && db <= se) || (de >= sb && de <= se);

...but this part I think the overlap calculation isn't right. Consider
db=5, de=15, sb=10, se=12. This gives overlap=false but
the two regions do overlap because [sb,se] is entirely inside [db,de]. I think you want
overlap = (db < se && sb < de);
(this is the same logic as each of the x/y range checks in the rectangle overlap test. put another way, if !(db<se) then we can't have an overlap because the dest range starts after the source range ends; similarly if !(sb<de) then the source range begins after the dest range ends and there's no overlap. So for an overlap to be possible we must have both db<se && sb<de.)
Here I'm using a definition of the "end" de and se which is that they point to the byte *after* the last one used (ie that we're really working with "half-open" ranges [db, de) and [sb, se) where de and se aren't in the range), because that's easier to calculate given that we need to account for bypp and it's more natural when dealing with "start, length" pairs.

Also and less importantly (because it's wrong in the "safe" direction) I think your se and de are overestimates, because one-past-the-last-used-byte in each case is
sb + (width + (height-1) * src_pitch) * bypp
(consider width=1 height=1, where one-past-the-last-used-byte is sb + bypp because there's only one pixel involved).

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 805

Quoted 7 lines
hw/display/sm501.c +4 -0
+            }
+            if (overlap) {
+                /* pixman can't do reverse blit: copy via temporary */
+                int tmp_stride = DIV_ROUND_UP(width * bypp, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
                 if (tmp_stride * sizeof(uint32_t) * height > sizeof(tmp_buf)) {

PS: why do we care about overlap only for right-to-left blits and not left-to-right blits ?

thanks
-- PMM

#11 lore raw

Re: [PATCH v3 9/9] sm501: Fix and optimize overlap check

Quoted 54 lines · Peter Maydell
Quoted 32 lines · BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

---
 hw/display/sm501.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
hw/display/sm501.c +1 -4
index 2db347dcbc..e7c69bf7fd 100644
@@ -690,6 +690,7 @@ static void sm501_2d_operation(SM501State *s)
     unsigned int dst_pitch = (s->twoD_pitch >> 16) & 0x1FFF;
     int crt = (s->dc_crt_control & SM501_DC_CRT_CONTROL_SEL) ? 1 : 0;
     int fb_len = get_width(s, crt) * get_height(s, crt) * get_bpp(s, crt);
+    bool overlap = false;
 
     if ((s->twoD_stretch >> 16) & 0xF) {
         qemu_log_mask(LOG_UNIMP, "sm501: only XY addressing is supported.\n");
@@ -784,16 +785,21 @@ static void sm501_2d_operation(SM501State *s)
                          ldn_he_p(&s->local_mem[src_base + si], bypp));
                 break;
             }
-            /* Check for overlaps, this could be made more exact */
-            uint32_t sb, se, db, de;
-            sb = src_base + src_x + src_y * (width + src_pitch);
-            se = sb + width + height * (width + src_pitch);

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 791

Quoted 10 lines
hw/display/sm501.c +4 -6
-            db = dst_base + dst_x + dst_y * (width + dst_pitch);
-            de = db + width + height * (width + dst_pitch);
-            if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
-                /* regions may overlap: copy via temporary */
-                int llb = width * bypp;
-                int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
+            /* If reverse blit do simple check for overlaps */
+            if (rtl && src_base == dst_base && src_pitch == dst_pitch) {
+                overlap = (src_x < dst_x + width && src_x + width > dst_x &&
+                           src_y < dst_y + height && src_y + height > dst_y);

This part looks good...

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 798

Quoted 7 lines
hw/display/sm501.c +7 -0
+            } else if (rtl) {
+                unsigned int sb, se, db, de;
+                sb = src_base + (src_x + src_y * src_pitch) * bypp;
+                se = sb + (width + height * src_pitch) * bypp;
+                db = dst_base + (dst_x + dst_y * dst_pitch) * bypp;
+                de = db + (width + height * dst_pitch) * bypp;
+                overlap = (db >= sb && db <= se) || (de >= sb && de <= se);
Quoted 11 lines

...but this part I think the overlap calculation isn't right. Consider db=5, de=15, sb=10, se=12. This gives overlap=false but
the two regions do overlap because [sb,se] is entirely inside [db,de]. I think you want
overlap = (db < se && sb < de);
(this is the same logic as each of the x/y range checks in the rectangle overlap test. put another way, if !(db<se) then we can't have an overlap because the dest range starts after the source range ends; similarly if !(sb<de) then the source range begins after the dest range ends and there's no overlap. So for an overlap to be possible we must have both db<se && sb<de.)

Thanks for checking it, you're right, I'll need to think about this again for a bit longer.

Quoted 6 lines

Here I'm using a definition of the "end" de and se which is that they point to the byte *after* the last one used (ie that we're really working with "half-open" ranges [db, de) and [sb, se) where de and se aren't in the range), because that's easier to calculate given that we need to account for bypp and it's more natural when dealing with "start, length" pairs.

Quoted 6 lines

Also and less importantly (because it's wrong in the "safe" direction) I think your se and de are overestimates, because one-past-the-last-used-byte in each case is
sb + (width + (height-1) * src_pitch) * bypp
(consider width=1 height=1, where one-past-the-last-used-byte is sb + bypp because there's only one pixel involved).

I'm not sure I follow the above but I had >= in the check to account for this although I've got the check wrong so that doesn't mean much.

What I've used is something like:

base pitch

+------------------------------------------+
|        ^                                 |
|      y |                                 |
|   x    v      w                          |
|< - - ->+-------------+                   |
|        |             |                   |
|       h|             |                   |
|        |             |                   |
|        +-------------+                   |
|                                          |
+------------------------------------------+

then y * pitch * bypp is the index of the line + x * bypp is index of first point which gives:

sb = base + (x + y * pitch) * bypp

then add width * bypp to get past the rect (index of byte after as you say above), then moved h lines down that's h * pitch * bypp but maybe that's too much because we only want to get to the last line not past that. Then using h - 1 is enough.

Quoted 10 lines

hw/display/sm501.c static void sm501_2d_operation(SM501State *s) · line 805

Quoted 7 lines
hw/display/sm501.c +4 -0
+            }
+            if (overlap) {
+                /* pixman can't do reverse blit: copy via temporary */
+                int tmp_stride = DIV_ROUND_UP(width * bypp, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
                 if (tmp_stride * sizeof(uint32_t) * height > sizeof(tmp_buf)) {

PS: why do we care about overlap only for right-to-left blits and not left-to-right blits ?

I think I've already answered that here:

https://lists.nongnu.org/archive/html/qemu-devel/2020-05/msg07650.html

since pixman does left to right that should match the intended operation even with overlap when rtl is not set. (Rtl also seems to mean bottom to top as well because there's only this one bit for direction.)

Regards,
BALATON Zoltan

#12 lore raw

PATCH v4 1/1 sm501: Fix and optimize overlap check

In reply to BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

---
This version fixes overlap check according to Peter's suggestion, only resending this patch as v4, others still valid from v3. Let me know if you want the whole series resent instead.

 hw/display/sm501.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
hw/display/sm501.c +16 -10
index 2db347dcbc..9cccc68c35 100644
@@ -690,6 +690,7 @@ static void sm501_2d_operation(SM501State *s)
     unsigned int dst_pitch = (s->twoD_pitch >> 16) & 0x1FFF;
     int crt = (s->dc_crt_control & SM501_DC_CRT_CONTROL_SEL) ? 1 : 0;
     int fb_len = get_width(s, crt) * get_height(s, crt) * get_bpp(s, crt);
+    bool overlap = false;
 
     if ((s->twoD_stretch >> 16) & 0xF) {
         qemu_log_mask(LOG_UNIMP, "sm501: only XY addressing is supported.\n");
@@ -784,16 +785,21 @@ static void sm501_2d_operation(SM501State *s)
                          ldn_he_p(&s->local_mem[src_base + si], bypp));
                 break;
             }
-            /* Check for overlaps, this could be made more exact */
-            uint32_t sb, se, db, de;
-            sb = src_base + src_x + src_y * (width + src_pitch);
-            se = sb + width + height * (width + src_pitch);
-            db = dst_base + dst_x + dst_y * (width + dst_pitch);
-            de = db + width + height * (width + dst_pitch);
-            if (rtl && ((db >= sb && db <= se) || (de >= sb && de <= se))) {
-                /* regions may overlap: copy via temporary */
-                int llb = width * bypp;
-                int tmp_stride = DIV_ROUND_UP(llb, sizeof(uint32_t));
+            /* If reverse blit do simple check for overlaps */
+            if (rtl && src_base == dst_base && src_pitch == dst_pitch) {
+                overlap = (src_x < dst_x + width && src_x + width > dst_x &&
+                           src_y < dst_y + height && src_y + height > dst_y);
+            } else if (rtl) {
+                unsigned int sb, se, db, de;
+                sb = src_base + (src_x + src_y * src_pitch) * bypp;
+                se = sb + (width + (height - 1) * src_pitch) * bypp;
+                db = dst_base + (dst_x + dst_y * dst_pitch) * bypp;
+                de = db + (width + (height - 1) * dst_pitch) * bypp;
+                overlap = (db < se && sb < de);
+            }
+            if (overlap) {
+                /* pixman can't do reverse blit: copy via temporary */
+                int tmp_stride = DIV_ROUND_UP(width * bypp, sizeof(uint32_t));
                 uint32_t *tmp = tmp_buf;
 
                 if (tmp_stride * sizeof(uint32_t) * height > sizeof(tmp_buf)) {
Signature · 1 line
2.21.3
Reviewed-by on v4 1/1
#13 lore raw

Re: [PATCH v4] sm501: Fix and optimize overlap check

Quoted 7 lines · BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

Quoted 5 lines

---
This version fixes overlap check according to Peter's suggestion, only resending this patch as v4, others still valid from v3. Let me know if you want the whole series resent instead.

thanks
-- PMM

#14 lore raw

Re: [PATCH v4] sm501: Fix and optimize overlap check

In reply to BALATON Zoltan
Quoted 11 lines · BALATON Zoltan

When doing reverse blit we need to check if source and dest overlap but it is not trivial due to possible different base and pitch of source and dest. Do rectangle overlap if base and pitch match, otherwise just check if memory area containing the rects overlaps so rects could possibly overlap.

---
This version fixes overlap check according to Peter's suggestion, only resending this patch as v4, others still valid from v3. Let me know if you want the whole series resent instead.

Queued up now, Last patch in a series is easy to replace.

thanks,
Gerd