From 6802c0593b10e090b21f7e70d72d5a63f3ac9251 Mon Sep 17 00:00:00 2001
From: TC1995 <mariogplayer90@gmail.com>
Date: Sat, 30 Sep 2023 22:08:08 +0200
Subject: [PATCH] Video, Storage and MCA changes/fixes. 1. Cirrus Logic GD54xx,
 Paradise/WD VGA now reset the interlace once a text mode is issued if not
 done automatically. 2. Paradise/WD's 15/16bpp modes using the 800x600
 resolution now have the correct ma_latch, should fix most operating systems
 drivers using this combo. 3. More fixes (hopefully) to the accelerated pitch
 and rowoffset of the Trident TGUI cards (9440AGi and 96x0XGi), should fix
 issues with delayed displays mode changes under various operating systems
 (e.g.: Win3.1x). 4. Preliminary implementation of the Area Fill command of
 XGA, which is issued while using various painting and/or calc utilities on
 Win3.1x (IBM XGA updated drivers, e.g.: 2.12). 5. Preliminary (and
 incomplete) 4bpp XGA mode. 6. The XGA memory test for the 0xa5 using writes
 (used by various operating systems) no longer conflicts with DOS' XGAKIT's
 memory detection. 7. Small ROP fixes to both XGA and 8514/A. 8. Re-organized
 the mapping of the Mach32 chipset, especially when to enable the ATI mode or
 switching back to IBM mode, should fix LFB conflicts with various operating
 systems. 9. According to The OS/2 Museum, the Adaptec AHA-154xB series of
 SCSI cards fail the ASPI4DOS.SYS 3.36 signature check, so now make the
 changes accordingly. 10. Remove useless and crashy bios-less option of the
 Trantor T128. 11. The Image Manager 1024 card can also be used on a XT
 (although only if it has a V20/V30). 12. Re-organized the IBM PS/2 model 60
 initialization as well as its right POS machine ID (though an update to
 sc.exe is still required for the POST memory amount to work normally).

---
 src/include/86box/scsi_x54x.h |   3 +-
 src/include/86box/vid_xga.h   |   3 +
 src/machine/m_ps2_mca.c       |  64 +++----
 src/scsi/scsi_ncr5380.c       |  29 ++-
 src/scsi/scsi_x54x.c          |  42 ++--
 src/video/vid_8514a.c         | 122 +++++++-----
 src/video/vid_ati_mach8.c     | 148 +++++++-------
 src/video/vid_cl54xx.c        |  11 ++
 src/video/vid_im1024.c        |   2 +-
 src/video/vid_paradise.c      |  25 ++-
 src/video/vid_svga.c          |  12 +-
 src/video/vid_tgui9440.c      |  52 ++++-
 src/video/vid_xga.c           | 349 +++++++++++++++++++++++++++-------
 13 files changed, 587 insertions(+), 275 deletions(-)

diff --git a/src/include/86box/scsi_x54x.h b/src/include/86box/scsi_x54x.h
index 22d4b35e4..0f1874302 100644
--- a/src/include/86box/scsi_x54x.h
+++ b/src/include/86box/scsi_x54x.h
@@ -366,6 +366,7 @@ typedef struct SGE_t {
 #define X54X_INT_GEOM_WRITABLE 8
 #define X54X_MBX_24BIT         16
 #define X54X_ISAPNP            32
+#define X54X_HAS_SIGNATURE     64
 
 typedef struct x54x_t {
     /* 32 bytes */
@@ -404,7 +405,7 @@ typedef struct x54x_t {
 
     /* for multi-threading, keep these volatile */
     volatile uint8_t Status;
-    volatile uint8_t Interrupt; 
+    volatile uint8_t Interrupt;
     volatile uint8_t MailboxIsBIOS;
     volatile uint8_t ToRaise;
     volatile uint8_t flags;
diff --git a/src/include/86box/vid_xga.h b/src/include/86box/vid_xga.h
index 97eb1456f..739e30c5c 100644
--- a/src/include/86box/vid_xga.h
+++ b/src/include/86box/vid_xga.h
@@ -201,6 +201,8 @@ typedef struct xga_t {
         int dst_map;
         int bkgd_src;
         int fore_src;
+        int oldx;
+        int oldy;
         int x;
         int y;
         int sx;
@@ -211,6 +213,7 @@ typedef struct xga_t {
         int py;
         int pattern;
         int command_len;
+        int filling;
 
         uint32_t short_stroke;
         uint32_t color_cmp;
diff --git a/src/machine/m_ps2_mca.c b/src/machine/m_ps2_mca.c
index b93f332db..a6fc30e1c 100644
--- a/src/machine/m_ps2_mca.c
+++ b/src/machine/m_ps2_mca.c
@@ -362,11 +362,6 @@ static void
 model_50_write(uint16_t port, uint8_t val)
 {
     switch (port) {
-        case 0x100:
-            ps2.io_id = val;
-            break;
-        case 0x101:
-            break;
         case 0x102:
             lpt1_remove();
             serial_remove(ps2.uart);
@@ -499,11 +494,6 @@ static void
 model_55sx_write(uint16_t port, uint8_t val)
 {
     switch (port) {
-        case 0x100:
-            ps2.io_id = val;
-            break;
-        case 0x101:
-            break;
         case 0x102:
             lpt1_remove();
             serial_remove(ps2.uart);
@@ -563,10 +553,6 @@ static void
 model_70_type3_write(uint16_t port, uint8_t val)
 {
     switch (port) {
-        case 0x100:
-            break;
-        case 0x101:
-            break;
         case 0x102:
             lpt1_remove();
             serial_remove(ps2.uart);
@@ -621,10 +607,6 @@ static void
 model_80_write(uint16_t port, uint8_t val)
 {
     switch (port) {
-        case 0x100:
-            break;
-        case 0x101:
-            break;
         case 0x102:
             lpt1_remove();
             serial_remove(ps2.uart);
@@ -985,12 +967,12 @@ ps2_mca_mem_d071_init(int start_mb)
 }
 
 static void
-ps2_mca_board_model_50_init(int slots)
+ps2_mca_board_model_50_init(void)
 {
     ps2_mca_board_common_init();
 
     mem_remap_top(384);
-    mca_init(slots);
+    mca_init(4);
     device_add(&keyboard_ps2_mca_2_device);
 
     ps2.planar_read  = model_50_read;
@@ -1005,6 +987,29 @@ ps2_mca_board_model_50_init(int slots)
         device_add(&ps1vga_mca_device);
 }
 
+static void
+ps2_mca_board_model_60_init(void)
+{
+    ps2_mca_board_common_init();
+
+    mem_remap_top(384);
+    mca_init(8);
+    device_add(&keyboard_ps2_mca_2_device);
+
+    ps2.planar_read  = model_50_read;
+    ps2.planar_write = model_50_write;
+
+    if (mem_size > 2048) {
+        /* Only 2 MB supported on planar, create a memory expansion card for the rest */
+        ps2_mca_mem_fffc_init(2);
+    }
+
+    device_add(&ps2_nvr_55ls_device);
+
+    if (gfxcard[0] == VID_INTERNAL)
+        device_add(&ps1vga_mca_device);
+}
+
 static void
 ps2_mca_board_model_55sx_init(int has_sec_nvram, int slots)
 {
@@ -1051,10 +1056,8 @@ ps2_mca_board_model_55sx_init(int has_sec_nvram, int slots)
     mca_init(slots);
     device_add(&keyboard_ps2_device);
 
-    if (has_sec_nvram == 1)
+    if (has_sec_nvram)
         device_add(&ps2_nvr_55ls_device);
-    else if (has_sec_nvram == 2)
-        device_add(&ps2_nvr_device);
 
     ps2.planar_read  = model_55sx_read;
     ps2.planar_write = model_55sx_write;
@@ -1312,7 +1315,7 @@ ps2_mca_board_model_70_type34_init(int is_type4, int slots)
 }
 
 static void
-ps2_mca_board_model_80_type2_init(int is486ps2)
+ps2_mca_board_model_80_type2_init(void)
 {
     ps2_mca_board_common_init();
 
@@ -1372,7 +1375,7 @@ ps2_mca_board_model_80_type2_init(int is486ps2)
                     NULL);
     mem_mapping_disable(&ps2.split_mapping);
 
-    if ((mem_size > 4096) && !is486ps2) {
+    if (mem_size > 4096) {
         /* Only 4 MB supported on planar, create a memory expansion card for the rest */
         if (mem_size > 12288)
             ps2_mca_mem_d071_init(4);
@@ -1426,7 +1429,7 @@ machine_ps2_model_50_init(const machine_t *model)
     machine_ps2_common_init(model);
 
     ps2.planar_id = 0xfbff;
-    ps2_mca_board_model_50_init(4);
+    ps2_mca_board_model_50_init();
 
     return ret;
 }
@@ -1448,8 +1451,8 @@ machine_ps2_model_60_init(const machine_t *model)
 
     machine_ps2_common_init(model);
 
-    ps2.planar_id = 0xfbff;
-    ps2_mca_board_model_50_init(8);
+    ps2.planar_id = 0xf7ff;
+    ps2_mca_board_model_60_init();
 
     return ret;
 }
@@ -1509,7 +1512,6 @@ machine_ps2_model_70_type3_init(const machine_t *model)
     machine_ps2_common_init(model);
 
     ps2.planar_id = 0xf9ff;
-
     ps2_mca_board_model_70_type34_init(0, 4);
 
     return ret;
@@ -1530,7 +1532,7 @@ machine_ps2_model_80_init(const machine_t *model)
     machine_ps2_common_init(model);
 
     ps2.planar_id = 0xfdff;
-    ps2_mca_board_model_80_type2_init(0);
+    ps2_mca_board_model_80_type2_init();
 
     return ret;
 }
@@ -1550,7 +1552,6 @@ machine_ps2_model_80_axx_init(const machine_t *model)
     machine_ps2_common_init(model);
 
     ps2.planar_id = 0xfff9;
-
     ps2_mca_board_model_70_type34_init(0, 8);
 
     return ret;
@@ -1571,7 +1572,6 @@ machine_ps2_model_70_type4_init(const machine_t *model)
     machine_ps2_common_init(model);
 
     ps2.planar_id = 0xf9ff;
-
     ps2_mca_board_model_70_type34_init(1, 4);
 
     return ret;
diff --git a/src/scsi/scsi_ncr5380.c b/src/scsi/scsi_ncr5380.c
index c4bec884d..656257849 100644
--- a/src/scsi/scsi_ncr5380.c
+++ b/src/scsi/scsi_ncr5380.c
@@ -1553,10 +1553,14 @@ ncr_init(const device_t *info)
                 ncr_dev->bios_ver = 1;
             }
 
-            if (ncr_dev->bios_ver == 1)
-                fn = RT1000B_820R_ROM;
-            else
-                fn = RT1000B_810R_ROM;
+            switch (ncr_dev->bios_ver) {
+                case 0:
+                    fn = RT1000B_810R_ROM;
+                    break;
+                case 1:
+                    fn = RT1000B_820R_ROM;
+                    break;
+            }
 
             rom_init(&ncr_dev->bios_rom, fn,
                      ncr_dev->rom_addr, 0x4000, 0x3fff, 0, MEM_MAPPING_EXTERNAL);
@@ -1601,9 +1605,8 @@ ncr_init(const device_t *info)
             ncr_dev->irq               = device_get_config_int("irq");
             ncr_dev->t128.bios_enabled = device_get_config_int("boot");
 
-            if (ncr_dev->t128.bios_enabled)
-                rom_init(&ncr_dev->bios_rom, T128_ROM,
-                         ncr_dev->rom_addr, 0x4000, 0x3fff, 0, MEM_MAPPING_EXTERNAL);
+            rom_init(&ncr_dev->bios_rom, T128_ROM,
+                     ncr_dev->rom_addr, 0x4000, 0x3fff, 0, MEM_MAPPING_EXTERNAL);
 
             mem_mapping_add(&ncr_dev->mapping, ncr_dev->rom_addr, 0x4000,
                             t128_read, NULL, NULL,
@@ -1635,15 +1638,12 @@ ncr_init(const device_t *info)
     ncr_log("%s\n", temp);
 
     ncr_reset(ncr_dev, &ncr_dev->ncr);
-    if (ncr_dev->type < 3 || ncr_dev->type == 4) {
+    if ((ncr_dev->type < 3) || (ncr_dev->type == 4)) {
         ncr_dev->status_ctrl     = STATUS_BUFFER_NOT_READY;
         ncr_dev->buffer_host_pos = 128;
     } else {
         ncr_dev->t128.status   = 0x04;
         ncr_dev->t128.host_pos = 512;
-
-        if (!ncr_dev->t128.bios_enabled)
-            ncr_dev->t128.status |= 0x80;
     }
     timer_add(&ncr_dev->timer, ncr_callback, ncr_dev, 0);
 
@@ -1894,13 +1894,6 @@ static const device_config_t t128_config[] = {
             { .description = ""                  }
         },
     },
-    {
-        .name = "boot",
-        .description = "Enable Boot ROM",
-        .type = CONFIG_BINARY,
-        .default_string = "",
-        .default_int = 1
-    },
     { .name = "", .description = "", .type = CONFIG_END }
 };
 // clang-format on
diff --git a/src/scsi/scsi_x54x.c b/src/scsi/scsi_x54x.c
index 2c714e18a..6876e4385 100644
--- a/src/scsi/scsi_x54x.c
+++ b/src/scsi/scsi_x54x.c
@@ -103,7 +103,7 @@ x54x_irq(x54x_t *dev, int set)
             else
                 picintc(1 << irq);
         }
-    }    
+    }
 }
 
 static void
@@ -466,6 +466,7 @@ x54x_bios_command(x54x_t *x54x, uint8_t max_id, BIOSCMD *cmd, int8_t islba)
                 }
 
                 return 0;
+                break;
 
             case 0x02: /* Read Desired Sectors to Memory */
             case 0x03: /* Write Desired Sectors from Memory */
@@ -1341,24 +1342,27 @@ x54x_in(uint16_t port, void *priv)
             if (dev->flags & X54X_INT_GEOM_WRITABLE)
                 ret = dev->Geometry;
             else {
-                switch (dev->Geometry) {
-                    default:
-                    case 0:
-                        ret = 'A';
-                        break;
-                    case 1:
-                        ret = 'D';
-                        break;
-                    case 2:
-                        ret = 'A';
-                        break;
-                    case 3:
-                        ret = 'P';
-                        break;
-                }
-                ret ^= 1;
-                dev->Geometry++;
-                dev->Geometry &= 0x03;
+                if (dev->flags & X54X_HAS_SIGNATURE) {
+                    switch (dev->Geometry) {
+                        default:
+                        case 0:
+                            ret = 'A';
+                            break;
+                        case 1:
+                            ret = 'D';
+                            break;
+                        case 2:
+                            ret = 'A';
+                            break;
+                        case 3:
+                            ret = 'P';
+                            break;
+                    }
+                    ret ^= 1;
+                    dev->Geometry++;
+                    dev->Geometry &= 0x03;
+                } else
+                    ret = 0xff;
                 break;
             }
             break;
diff --git a/src/video/vid_8514a.c b/src/video/vid_8514a.c
index 8ee0daadd..ef735852c 100644
--- a/src/video/vid_8514a.c
+++ b/src/video/vid_8514a.c
@@ -144,13 +144,13 @@ ibm8514_log(const char *fmt, ...)
                 dest_dat = MAX(src_dat, dest_dat);                                            \
                 break;                                                                        \
             case 0x15:                                                                        \
-                dest_dat = (dest_dat - src_dat) / 2;                                          \
+                dest_dat = (dest_dat - src_dat) >> 1;                                         \
                 break;                                                                        \
             case 0x16:                                                                        \
-                dest_dat = (src_dat - dest_dat) / 2;                                          \
+                dest_dat = (src_dat - dest_dat) >> 1;                                         \
                 break;                                                                        \
             case 0x17:                                                                        \
-                dest_dat = (dest_dat + src_dat) / 2;                                          \
+                dest_dat = (dest_dat + src_dat) >> 1;                                         \
                 break;                                                                        \
             case 0x18:                                                                        \
                 dest_dat = MAX(0, (dest_dat - src_dat));                                      \
@@ -162,7 +162,7 @@ ibm8514_log(const char *fmt, ...)
                 dest_dat = MAX(0, (src_dat - dest_dat));                                      \
                 break;                                                                        \
             case 0x1b:                                                                        \
-                dest_dat = MIN(0xff, (dest_dat + src_dat));                                   \
+                dest_dat = MIN(~0, (dest_dat + src_dat));                                     \
                 break;                                                                        \
             case 0x1c:                                                                        \
                 dest_dat = MAX(0, (dest_dat - src_dat)) / 2;                                  \
@@ -174,7 +174,7 @@ ibm8514_log(const char *fmt, ...)
                 dest_dat = MAX(0, (src_dat - dest_dat)) / 2;                                  \
                 break;                                                                        \
             case 0x1f:                                                                        \
-                dest_dat = (0xff < (src_dat + dest_dat)) ? 0xff : ((src_dat + dest_dat) / 2); \
+                dest_dat = (~0 < (src_dat + dest_dat)) ? ~0 : ((src_dat + dest_dat) >> 1); 	  \
                 break;                                                                        \
         }                                                                                     \
     }
@@ -1297,7 +1297,7 @@ ibm8514_accel_start(int count, int cpu_input, uint32_t mix_dat, uint32_t cpu_dat
     /*Bit 4 of the Command register is the draw yes bit, which enables writing to memory/reading from memory when enabled.
       When this bit is disabled, no writing to memory/reading from memory is allowed. (This bit is almost meaningless on
       the NOP command)*/
-    ibm8514_log("CMD8514: CMD=%d, full=%04x, pixcntl=%x, count=%d.\n", cmd, dev->accel.cmd, pixcntl, count);
+    ibm8514_log("CMD8514: CMD=%d, full=%04x, pixcntl=%x, count=%d, frgdmix = %02x, bkgdmix = %02x, polygon=%x.\n", cmd, dev->accel.cmd, pixcntl, count, frgd_mix, bkgd_mix, dev->accel.multifunc[0x0a] & 6);
 
     switch (cmd) {
         case 0: /*NOP (Short Stroke Vectors)*/
@@ -3057,13 +3057,39 @@ rect_fill:
                         old_dest_dat = dest_dat;
                         MIX(mix_dat & mix_mask, dest_dat, src_dat);
                         dest_dat = (dest_dat & wrt_mask) | (old_dest_dat & ~wrt_mask);
-                        if ((dev->accel.cmd & 4) && (dev->accel.sy < dev->accel.maj_axis_pcnt)) {
-                            if (!dev->accel.sy) {
-                                WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
-                            } else if ((dev->accel.cmd & 0x40) && dev->accel.sy && (dev->accel.cy == (dev->accel.oldcy + 1))) {
-                                WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
-                            } else if (!(dev->accel.cmd & 0x40) && dev->accel.sy && (dev->accel.err_term >= 0) && (dev->accel.cy == (dev->accel.oldcy + 1))) {
+                        if (dev->accel.cmd & 4) {
+                            if (dev->accel.sy < dev->accel.maj_axis_pcnt) {
+                                if (dev->accel.cmd & 0x40) {
+                                    WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                                } else {
+                                    if (dev->accel.cy == (dev->accel.oldcy + 1)) {
+                                        if (dev->accel.cmd & 0x20) {
+                                            if (dev->accel.err_term < (dev->accel.destx_distp + dev->accel.desty_axstp)) {
+                                                WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                                            }
+                                        } else {
+                                            if (dev->accel.err_term >= 0) {
+                                                WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                                            }
+                                        }
+                                    }
+                                }
+                            }
+                        } else {
+                            if (dev->accel.cmd & 0x40) {
                                 WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                            } else {
+                                if (dev->accel.cy == (dev->accel.oldcy + 1)) {
+                                    if (dev->accel.cmd & 0x20) {
+                                        if (dev->accel.err_term < (dev->accel.destx_distp + dev->accel.desty_axstp)) {
+                                            WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                                        }
+                                    } else {
+                                        if (dev->accel.err_term >= 0) {
+                                            WRITE((dev->accel.cy * dev->pitch) + (dev->accel.cx), dest_dat);
+                                        }
+                                    }
+                                }
                             }
                         }
                     }
@@ -3080,7 +3106,42 @@ rect_fill:
                     break;
                 }
 
-                if (dev->accel.err_term >= dev->accel.maj_axis_pcnt) {
+                /*Step major axis*/
+                switch (dev->accel.cmd & 0xe0) {
+                    case 0x00:
+                        dev->accel.cx--;
+                        break;
+                    case 0x20:
+                        dev->accel.cx++;
+                        break;
+                    case 0x40:
+                        dev->accel.oldcy = dev->accel.cy;
+                        dev->accel.cy--;
+                        break;
+                    case 0x60:
+                        dev->accel.oldcy = dev->accel.cy;
+                        dev->accel.cy--;
+                        break;
+                    case 0x80:
+                        dev->accel.cx--;
+                        break;
+                    case 0xa0:
+                        dev->accel.cx++;
+                        break;
+                    case 0xc0:
+                        dev->accel.oldcy = dev->accel.cy;
+                        dev->accel.cy++;
+                        break;
+                    case 0xe0:
+                        dev->accel.oldcy = dev->accel.cy;
+                        dev->accel.cy++;
+                        break;
+
+                    default:
+                        break;
+                }
+
+                if (dev->accel.err_term >= 0) {
                     dev->accel.err_term += dev->accel.destx_distp;
                     /*Step minor axis*/
                     switch (dev->accel.cmd & 0xe0) {
@@ -3119,41 +3180,6 @@ rect_fill:
                 } else
                     dev->accel.err_term += dev->accel.desty_axstp;
 
-                /*Step major axis*/
-                switch (dev->accel.cmd & 0xe0) {
-                    case 0x00:
-                        dev->accel.cx--;
-                        break;
-                    case 0x20:
-                        dev->accel.cx++;
-                        break;
-                    case 0x40:
-                        dev->accel.oldcy = dev->accel.cy;
-                        dev->accel.cy--;
-                        break;
-                    case 0x60:
-                        dev->accel.oldcy = dev->accel.cy;
-                        dev->accel.cy--;
-                        break;
-                    case 0x80:
-                        dev->accel.cx--;
-                        break;
-                    case 0xa0:
-                        dev->accel.cx++;
-                        break;
-                    case 0xc0:
-                        dev->accel.oldcy = dev->accel.cy;
-                        dev->accel.cy++;
-                        break;
-                    case 0xe0:
-                        dev->accel.oldcy = dev->accel.cy;
-                        dev->accel.cy++;
-                        break;
-
-                    default:
-                        break;
-                }
-
                 dev->accel.sy++;
             }
             break;
diff --git a/src/video/vid_ati_mach8.c b/src/video/vid_ati_mach8.c
index 8df054060..1d3f6bbc8 100644
--- a/src/video/vid_ati_mach8.c
+++ b/src/video/vid_ati_mach8.c
@@ -99,6 +99,7 @@ typedef struct mach_t {
     uint8_t  bank_w;
     uint8_t  bank_r;
     uint16_t shadow_set;
+    int ext_on;
 
     struct {
         uint8_t  line_idx;
@@ -2599,7 +2600,6 @@ mach_recalctimings(svga_t *svga)
 
     if (dev->on) {
         mach_log("8514/A ON.\n");
-        svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock((mach->accel.clock_sel >> 2) & 0x0f, svga->clock_gen);
         if (dev->local >= 2) {
             dev->h_disp                     = (dev->hdisp + 1) << 3;
             dev->h_total                    = (dev->htotal + 1);
@@ -2614,6 +2614,9 @@ mach_recalctimings(svga_t *svga)
             if (dev->dispend == 598)
                 dev->dispend += 2;
 
+            if (dev->h_disp == 1024)
+                dev->accel.advfunc_cntl |= 4; /*Bit 2 means high resolution e.g.: 1024x768*/
+
             if (dev->accel.advfunc_cntl & 4) {
                 if (mach->shadow_set & 2) {
                     if (dev->h_disp == 8) {
@@ -2622,14 +2625,20 @@ mach_recalctimings(svga_t *svga)
                         dev->v_total = 1536;
                         dev->v_syncstart = 1536;
                     }
-                }
+                    mach_log("Shadow set 2.\n");
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / 44900000.0;
+                } else
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock((mach->accel.clock_sel >> 2) & 0x0f, svga->clock_gen);
             } else {
-                if (mach->shadow_set & 1) {
-                    if (dev->h_disp == 1024) {
-                        dev->h_disp = 640;
-                        dev->dispend = 480;
-                    }
+                if (dev->h_disp == 1024) {
+                    dev->h_disp = 640;
+                    dev->dispend = 480;
                 }
+                if (mach->shadow_set & 1) {
+                    mach_log("Shadow set 1.\n");
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / 25175000.0;
+                } else
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock((mach->accel.clock_sel >> 2) & 0x0f, svga->clock_gen);
             }
 
             if (dev->interlace) {
@@ -2721,14 +2730,18 @@ mach_recalctimings(svga_t *svga)
                         dev->v_total = 1536;
                         dev->v_syncstart = 1536;
                     }
-                }
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / 44900000.0;
+                } else
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock((mach->accel.clock_sel >> 2) & 0x0f, svga->clock_gen);
             } else {
-                if (mach->shadow_set & 1) {
-                    if (dev->h_disp == 1024) {
-                        dev->h_disp = 640;
-                        dev->dispend = 480;
-                    }
+                if (dev->h_disp == 1024) {
+                    dev->h_disp = 640;
+                    dev->dispend = 480;
                 }
+                if (mach->shadow_set & 1) {
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / 25175000.0;
+                } else
+                    svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock((mach->accel.clock_sel >> 2) & 0x0f, svga->clock_gen);
             }
             if (dev->interlace) {
                 dev->dispend >>= 1;
@@ -2740,7 +2753,7 @@ mach_recalctimings(svga_t *svga)
             }
             dev->pitch = dev->ext_pitch;
             dev->rowoffset = dev->ext_crt_pitch;
-            mach_log("hv(%d,%d), pitch=%d, rowoffset=%d, gextconfig=%03x.\n", dev->h_disp, dev->dispend, dev->pitch, dev->rowoffset, mach->accel.ext_ge_config & 0xcec0);
+            mach_log("cntl=%d, hv(%d,%d), pitch=%d, rowoffset=%d, gextconfig=%03x.\n", dev->accel.advfunc_cntl & 4, dev->h_disp, dev->dispend, dev->pitch, dev->rowoffset, mach->accel.ext_ge_config & 0xcec0);
             svga->map8 = dev->pallook;
             svga->render8514 = ibm8514_render_8bpp;
             if (mach->regs[0xb8] & 0x40)
@@ -2750,6 +2763,7 @@ mach_recalctimings(svga_t *svga)
 
     if (!svga->scrblank && (svga->crtc[0x17] & 0x80) && svga->attr_palette_enable) {
         if (((svga->gdcreg[6] & 1) || (svga->attrregs[0x10] & 1))) {
+            mach_log("VGA clock=%02x.\n", mach->regs[0xa7] & 0x80);
             svga->clock = (cpuclock * (double) (1ULL << 32)) / svga->getclock(clock_sel, svga->clock_gen);
             if (mach->regs[0xa7] & 0x80)
                 svga->clock *= 3;
@@ -3688,7 +3702,9 @@ mach_accel_out(uint16_t port, uint8_t val, mach_t *mach)
             } else {
                 dev->on = (dev->accel.advfunc_cntl & 0x01);
                 vga_on = !dev->on;
-                mach_log("ATI 8514/A: (0x4ae8) val = %04x\n", val & 0x01);
+                mach->ext_on = dev->on;
+                mach_log("ATI 8514/A: (0x4ae9) val = %04x, ext = %d.\n", dev->accel.advfunc_cntl & 0x01, mach->ext_on);
+                mach32_updatemapping(mach);
             }
             svga_recalctimings(svga);
             break;
@@ -3777,6 +3793,7 @@ mach_accel_out(uint16_t port, uint8_t val, mach_t *mach)
 
         case 0x36ee:
         case 0x36ef:
+            mach_log("ATI 8514/A: (0x%04x) val = %04x.\n", port, val);
             WRITE8(port, mach->misc, val);
             mach->misc &= 0xfff0;
             break;
@@ -3813,21 +3830,27 @@ mach_accel_out(uint16_t port, uint8_t val, mach_t *mach)
         case 0x4aef:
             WRITE8(port, mach->accel.clock_sel, val);
             if (port & 1) {
-                mach_log("ATI 8514/A: (0x4aee) val = %04x\n", mach->accel.clock_sel & 0x01);
                 dev->on = mach->accel.clock_sel & 0x01;
                 vga_on = !dev->on;
+                pclog("ATI 8514/A: (0x4aef) val = %04x, ext = %d.\n", mach->accel.clock_sel & 0x01, mach->ext_on);
             }
             svga_recalctimings(svga);
             break;
 
         case 0x52ee:
         case 0x52ef:
+            mach_log("ATI 8514/A: (0x%04x) val = %04x.\n", port, val);
             WRITE8(port, mach->accel.scratch0, val);
+            if (port & 1)
+                mach->ext_on = 1;
             break;
 
         case 0x56ee:
         case 0x56ef:
+            mach_log("ATI 8514/A: (0x%04x) val = %04x.\n", port, val);
             WRITE8(port, mach->accel.scratch1, val);
+            if (port & 1)
+                mach->ext_on = 1;
             break;
 
         case 0x5aee:
@@ -4744,13 +4767,8 @@ mach32_write_linear(uint32_t addr, uint8_t val, void *priv)
 static void
 mach32_write(uint32_t addr, uint8_t val, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
-
-    if (!dev->on) {
-        svga_write(addr, val, svga);
-        return;
-    }
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
 
     addr = (addr & svga->banked_mask) + svga->write_bank;
     mach32_write_linear(addr, val, svga);
@@ -4759,33 +4777,25 @@ mach32_write(uint32_t addr, uint8_t val, void *priv)
 static void
 mach32_writew(uint32_t addr, uint16_t val, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
 
-    if (!dev->on) {
-        svga_writew(addr, val, svga);
-        return;
-    }
-
-    mach32_write(addr, val & 0xff, svga);
-    mach32_write(addr + 1, val >> 8, svga);
+    addr = (addr & svga->banked_mask) + svga->write_bank;
+    mach32_write_linear(addr, val & 0xff, svga);
+    mach32_write_linear(addr + 1, val >> 8, svga);
 }
 
 static void
 mach32_writel(uint32_t addr, uint32_t val, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
 
-    if (!dev->on) {
-        svga_writel(addr, val, svga);
-        return;
-    }
-
-    mach32_write(addr, val & 0xff, svga);
-    mach32_write(addr + 1, val >> 8, svga);
-    mach32_write(addr + 2, val >> 16, svga);
-    mach32_write(addr + 3, val >> 24, svga);
+    addr = (addr & svga->banked_mask) + svga->write_bank;
+    mach32_write_linear(addr, val & 0xff, svga);
+    mach32_write_linear(addr + 1, val >> 8, svga);
+    mach32_write_linear(addr + 2, val >> 16, svga);
+    mach32_write_linear(addr + 3, val >> 24, svga);
 }
 
 static uint8_t
@@ -4872,15 +4882,10 @@ mach32_read_linear(uint32_t addr, void *priv)
 static uint8_t
 mach32_read(uint32_t addr, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
     uint8_t          ret;
 
-    if (!dev->on) {
-        ret = svga_read(addr, svga);
-        return ret;
-    }
-
     addr = (addr & svga->banked_mask) + svga->read_bank;
     ret  = mach32_read_linear(addr, svga);
     return ret;
@@ -4889,36 +4894,28 @@ mach32_read(uint32_t addr, void *priv)
 static uint16_t
 mach32_readw(uint32_t addr, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
     uint16_t         ret;
 
-    if (!dev->on) {
-        ret = svga_readw(addr, svga);
-        return ret;
-    }
-
-    ret  = mach32_read(addr, svga);
-    ret  |= (mach32_read(addr + 1, svga) << 8);
+    addr = (addr & svga->banked_mask) + svga->read_bank;
+    ret  = mach32_read_linear(addr, svga);
+    ret  |= (mach32_read_linear(addr + 1, svga) << 8);
     return ret;
 }
 
 static uint32_t
 mach32_readl(uint32_t addr, void *priv)
 {
-    svga_t          *svga = (svga_t *) priv;
-    const ibm8514_t *dev  = &svga->dev8514;
+    mach_t *mach = (mach_t *) priv;
+    svga_t          *svga = &mach->svga;
     uint32_t         ret;
 
-    if (!dev->on) {
-        ret = svga_readl(addr, svga);
-        return ret;
-    }
-
-    ret  = mach32_read(addr, svga);
-    ret  |= (mach32_read(addr + 1, svga) << 8);
-    ret  |= (mach32_read(addr + 2, svga) << 16);
-    ret  |= (mach32_read(addr + 3, svga) << 24);
+    addr = (addr & svga->banked_mask) + svga->read_bank;
+    ret  = mach32_read_linear(addr, svga);
+    ret  |= (mach32_read_linear(addr + 1, svga) << 8);
+    ret  |= (mach32_read_linear(addr + 2, svga) << 16);
+    ret  |= (mach32_read_linear(addr + 3, svga) << 24);
     return ret;
 }
 
@@ -4986,7 +4983,7 @@ mach32_ap_writel(uint32_t addr, uint32_t val, void *priv)
             mach_accel_outw(0x02e8 + (port_dword << 8) + 4, val >> 16, mach);
         }
     } else {
-        mach_log("Linear WORDL Write=%08x, val=%08x, mode=%d, rop=%02x.\n", addr, val, mach->svga.writemode, mach->svga.gdcreg[3] & 0x18);
+        mach_log("Linear WORDL Write=%08x.\n", addr);
         mach32_write_linear(addr, val & 0xff, svga);
         mach32_write_linear(addr + 1, val >> 8, svga);
         mach32_write_linear(addr + 2, val >> 16, svga);
@@ -5069,6 +5066,7 @@ static void
 mach32_updatemapping(mach_t *mach)
 {
     svga_t *svga = &mach->svga;
+    ibm8514_t *dev = &svga->dev8514;
 
     if (mach->pci_bus && (!(mach->pci_regs[PCI_REG_COMMAND] & PCI_COMMAND_MEM))) {
         mem_mapping_disable(&svga->mapping);
@@ -5118,6 +5116,13 @@ mach32_updatemapping(mach_t *mach)
         mach->ap_size = 4;
         mem_mapping_disable(&mach->mmio_linear_mapping);
     }
+    if (mach->ext_on && (dev->local >= 2)) {
+        mem_mapping_set_handler(&svga->mapping, mach32_read, mach32_readw, mach32_readl, mach32_write, mach32_writew, mach32_writel);
+        mem_mapping_set_p(&svga->mapping, mach);
+    } else {
+        mem_mapping_set_handler(&svga->mapping, svga_read, svga_readw, svga_readl, svga_write, svga_writew, svga_writel);
+        mem_mapping_set_p(&svga->mapping, svga);
+    }
 }
 
 static void
@@ -5692,7 +5697,6 @@ mach8_init(const device_t *info)
             mach->config1 |= 0x0400;
             svga->clock_gen = device_add(&ati18811_0_device);
         }
-        mem_mapping_set_handler(&svga->mapping, mach32_read, mach32_readw, mach32_readl, mach32_write, mach32_writew, mach32_writel);
         mem_mapping_add(&mach->mmio_linear_mapping, 0, 0, mach32_ap_readb, mach32_ap_readw, mach32_ap_readl, mach32_ap_writeb, mach32_ap_writew, mach32_ap_writel, NULL, MEM_MAPPING_EXTERNAL, mach);
         mem_mapping_disable(&mach->mmio_linear_mapping);
     } else {
diff --git a/src/video/vid_cl54xx.c b/src/video/vid_cl54xx.c
index f5742e6d0..41aadf6ba 100644
--- a/src/video/vid_cl54xx.c
+++ b/src/video/vid_cl54xx.c
@@ -1739,6 +1739,10 @@ gd54xx_recalctimings(svga_t *svga)
 
     svga->interlace = (svga->crtc[0x1a] & 0x01);
 
+    if (!(svga->gdcreg[6] & 1) && !(svga->attrregs[0x10] & 1)) { /*Text mode*/
+        svga->interlace = 0;
+    }
+
     svga->map8 = svga->pallook;
     if (svga->seqregs[7] & CIRRUS_SR7_BPP_SVGA) {
         if (linedbl)
@@ -1921,6 +1925,13 @@ gd54xx_recalctimings(svga_t *svga)
     }
 
     svga->vram_display_mask = (svga->crtc[0x1b] & 2) ? gd54xx->vram_mask : 0x3ffff;
+
+    if (!(svga->gdcreg[6] & 1) && !(svga->attrregs[0x10] & 1)) { /*Text mode*/
+        if (svga->seqregs[1] & 8) {
+            svga->render = svga_render_text_40;
+        } else
+            svga->render = svga_render_text_80;
+    }
 }
 
 static void
diff --git a/src/video/vid_im1024.c b/src/video/vid_im1024.c
index 1e60a3258..c7602ffd0 100644
--- a/src/video/vid_im1024.c
+++ b/src/video/vid_im1024.c
@@ -1085,7 +1085,7 @@ im1024_speed_changed(void *priv)
 const device_t im1024_device = {
     .name          = "ImageManager 1024",
     .internal_name = "im1024",
-    .flags         = DEVICE_ISA | DEVICE_AT,
+    .flags         = DEVICE_ISA,
     .local         = 0,
     .init          = im1024_init,
     .close         = im1024_close,
diff --git a/src/video/vid_paradise.c b/src/video/vid_paradise.c
index a98753337..ad197f302 100644
--- a/src/video/vid_paradise.c
+++ b/src/video/vid_paradise.c
@@ -341,23 +341,44 @@ paradise_recalctimings(svga_t *svga)
         }
     }
 
+    if (!(svga->gdcreg[6] & 1) && !(svga->attrregs[0x10] & 1)) { /*Text mode*/
+        svga->interlace = 0;
+    }
+
     if (paradise->type < WD90C30) {
-        if (svga->bpp >= 8 && !svga->lowres) {
+        if ((svga->bpp >= 8) && !svga->lowres) {
             svga->render = svga_render_8bpp_highres;
         }
     } else {
-        if (svga->bpp >= 8 && !svga->lowres) {
+        if ((svga->bpp >= 8) && !svga->lowres) {
             if (svga->bpp == 16) {
                 svga->render = svga_render_16bpp_highres;
                 svga->hdisp >>= 1;
+                if (svga->hdisp == 788)
+                    svga->hdisp += 12;
+                if (svga->hdisp == 800)
+                    svga->ma_latch -= 3;
             } else if (svga->bpp == 15) {
                 svga->render = svga_render_15bpp_highres;
                 svga->hdisp >>= 1;
+                if (svga->hdisp == 788)
+                    svga->hdisp += 12;
+                if (svga->hdisp == 800)
+                    svga->ma_latch -= 3;
             } else {
                 svga->render = svga_render_8bpp_highres;
             }
         }
     }
+
+    if (!(svga->gdcreg[6] & 1) && !(svga->attrregs[0x10] & 1)) { /*Text mode*/
+        if (svga->hdisp == 360)
+            svga->hdisp <<= 1;
+        if (svga->seqregs[1] & 8) {
+            svga->render = svga_render_text_40;
+        } else
+            svga->render = svga_render_text_80;
+    }
 }
 
 static void
diff --git a/src/video/vid_svga.c b/src/video/vid_svga.c
index 6a9f2099b..e32575db7 100644
--- a/src/video/vid_svga.c
+++ b/src/video/vid_svga.c
@@ -830,7 +830,7 @@ svga_poll(void *priv)
         return;
     }
     if (xga_active && xga->on) {
-        if ((xga->disp_cntl_2 & 7) >= 3) {
+        if ((xga->disp_cntl_2 & 7) >= 2) {
             xga_poll(xga, svga);
             return;
         }
@@ -1247,11 +1247,12 @@ svga_write_common(uint32_t addr, uint8_t val, uint8_t linear, void *priv)
             if (((xga->op_mode & 7) >= 4) && (xga->aperture_cntl >= 1)) {
                 if (val == 0xa5) { /*Memory size test of XGA*/
                     xga->test    = val;
-                    xga->a5_test = 1;
+                    if (addr == 0xa0001)
+                        xga->a5_test = 1;
+
                     xga->on = 0;
                     vga_on = 1;
                     xga->disp_cntl_2 = 0;
-                    xga->clk_sel_1 = 0;
                     svga_log("XGA test1 addr = %05x.\n", addr);
                     return;
                 } else if (val == 0x5a) {
@@ -1259,7 +1260,7 @@ svga_write_common(uint32_t addr, uint8_t val, uint8_t linear, void *priv)
                     xga->on = 0;
                     vga_on = 1;
                     xga->disp_cntl_2 = 0;
-                    xga->clk_sel_1 = 0;
+                    svga_log("XGA test2 addr = %05x.\n", addr);
                     return;
                 } else if ((addr == 0xa0000) || (addr == 0xa0010)) {
                     addr += xga->write_bank;
@@ -1469,6 +1470,7 @@ svga_read_common(uint32_t addr, uint8_t linear, void *priv)
                         xga->on = 1;
                         vga_on = 0;
                     } else if ((addr == 0xa0000) && xga->a5_test) { /*This is required by XGAKIT to pass the memory test*/
+                        svga_log("A5 test bank = %x.\n", addr);
                         addr += xga->read_bank;
                         ret = xga->vram[addr & xga->vram_mask];
                     } else {
@@ -1476,7 +1478,7 @@ svga_read_common(uint32_t addr, uint8_t linear, void *priv)
                         xga->on = 1;
                         vga_on = 0;
                     }
-                    svga_log("A5 read: XGA ON = %d, addr = %05x.\n", xga->on, addr);
+                    svga_log("A5 read: XGA ON = %d, addr = %05x, ret = %02x, test1 = %x.\n", xga->on, addr, ret, xga->a5_test);
                     return ret;
                 } else if (xga->test == 0x5a) {
                     ret = xga->test;
diff --git a/src/video/vid_tgui9440.c b/src/video/vid_tgui9440.c
index 9204e8644..0f9473d1a 100644
--- a/src/video/vid_tgui9440.c
+++ b/src/video/vid_tgui9440.c
@@ -1399,6 +1399,7 @@ tgui_accel_command(int count, uint32_t cpu_dat, tgui_t *tgui)
     uint32_t        trans_col = (tgui->accel.flags & TGUI_TRANSREV) ? tgui->accel.fg_col : tgui->accel.bg_col;
     uint16_t       *vram_w    = (uint16_t *) svga->vram;
     uint32_t       *vram_l    = (uint32_t *) svga->vram;
+    uint8_t   		ger22upper = (tgui->accel.ger22 >> 8);
 
     if (tgui->accel.bpp == 0) {
         trans_col &= 0xff;
@@ -1454,6 +1455,16 @@ tgui_accel_command(int count, uint32_t cpu_dat, tgui_t *tgui)
         }
     }
 
+    /*For delayed mode switches.*/
+    if (tgui->type == TGUI_9440) {
+        if (tgui->accel.pitch == 800)
+            tgui->accel.pitch += 32;
+
+        if (tgui->accel.bpp == 1) {
+            if (!ger22upper)
+                tgui->accel.pitch = svga->rowoffset << 2;
+        }
+    }
 #if 0
     pclog("TGUI accel command = %x, ger22 = %04x, hdisp = %d, dispend = %d, vtotal = %d, rowoffset = %d, svgabpp = %d, interlace = %d, accelbpp = %d, pitch = %d.\n", tgui->accel.command, tgui->accel.ger22, svga->hdisp, svga->dispend, svga->vtotal, svga->rowoffset, svga->bpp, svga->interlace, tgui->accel.bpp, tgui->accel.pitch);
 #endif
@@ -2066,6 +2077,19 @@ tgui_accel_out(uint16_t addr, uint8_t val, void *priv)
 
         case 0x2123:
             //pclog("Pitch IO23: val = %02x, rowoffset = %x, pitch = %d.\n", val, svga->rowoffset, tgui->accel.pitch);
+            switch (svga->bpp) {
+                case 8:
+                case 24:
+                    tgui->accel.bpp = 0;
+                    break;
+                case 15:
+                case 16:
+                    tgui->accel.bpp = 1;
+                    break;
+                case 32:
+                    tgui->accel.bpp = 3;
+                    break;
+            }
             tgui->accel.ger22 = (tgui->accel.ger22 & 0xff) | (val << 8);
             if ((val & 0x80) || ((val & 0xc0) == 0x40))
                 tgui->accel.pitch = svga->rowoffset << 3;
@@ -2075,15 +2099,14 @@ tgui_accel_out(uint16_t addr, uint8_t val, void *priv)
                     tgui->accel.pitch = 1024;
             }
 
+            if (tgui->accel.pitch == 800)
+                tgui->accel.pitch += 32;
+
             if (tgui->accel.bpp == 1)
                 tgui->accel.pitch >>= 1;
             else if (tgui->accel.bpp == 3)
                 tgui->accel.pitch >>= 2;
 
-
-            if (tgui->accel.pitch == 800)
-                tgui->accel.pitch += 32;
-
             svga_recalctimings(svga);
             break;
 
@@ -2722,7 +2745,19 @@ tgui_accel_write(uint32_t addr, uint8_t val, void *priv)
             break;
 
         case 0x23:
-            //pclog("Pitch MM23: val = %02x, rowoffset = %x, pitch = %d.\n", val, svga->rowoffset, tgui->accel.pitch);
+            switch (svga->bpp) {
+                case 8:
+                case 24:
+                    tgui->accel.bpp = 0;
+                    break;
+                case 15:
+                case 16:
+                    tgui->accel.bpp = 1;
+                    break;
+                case 32:
+                    tgui->accel.bpp = 3;
+                    break;
+            }
             tgui->accel.ger22 = (tgui->accel.ger22 & 0xff) | (val << 8);
             if ((val & 0x80) || ((val & 0xc0) == 0x40))
                 tgui->accel.pitch = svga->rowoffset << 3;
@@ -2732,14 +2767,15 @@ tgui_accel_write(uint32_t addr, uint8_t val, void *priv)
                     tgui->accel.pitch = 1024;
             }
 
+            if (tgui->accel.pitch == 800)
+                tgui->accel.pitch += 32;
+
             if (tgui->accel.bpp == 1)
                 tgui->accel.pitch >>= 1;
             else if (tgui->accel.bpp == 3)
                 tgui->accel.pitch >>= 2;
 
-            if (tgui->accel.pitch == 800)
-                tgui->accel.pitch += 32;
-
+            //pclog("Pitch MM23: fullval = %04x, rowoffset = %x, pitch = %d.\n", tgui->accel.ger22, svga->rowoffset, tgui->accel.pitch);
             svga_recalctimings(svga);
             break;
 
diff --git a/src/video/vid_xga.c b/src/video/vid_xga.c
index 1747bb13c..2c1749402 100644
--- a/src/video/vid_xga.c
+++ b/src/video/vid_xga.c
@@ -49,6 +49,7 @@ static uint8_t xga_ext_inb(uint16_t addr, void *priv);
 static void     xga_writew(uint32_t addr, uint16_t val, void *priv);
 static uint16_t xga_readw(uint32_t addr, void *priv);
 
+static void xga_render_4bpp(svga_t *svga);
 static void xga_render_8bpp(svga_t *svga);
 static void xga_render_16bpp(svga_t *svga);
 
@@ -376,7 +377,7 @@ xga_ext_out_reg(xga_t *xga, svga_t *svga, uint8_t idx, uint8_t val)
         case 0x51:
             xga_log("Reg51 write = %02x.\n", val);
             xga->disp_cntl_2 = val;
-            xga->on          = ((val & 7) >= 3);
+            xga->on          = ((val & 7) >= 2);
             vga_on           = !xga->on;
             svga_recalctimings(svga);
             break;
@@ -406,7 +407,7 @@ xga_ext_out_reg(xga_t *xga, svga_t *svga, uint8_t idx, uint8_t val)
             if ((xga->sprite_pos >= 0) && (xga->sprite_pos <= 16)) {
                 if ((xga->op_mode & 7) >= 5)
                     xga->cursor_data_on = 1;
-                else if ((xga->sprite_pos >= 1) || ((xga->disp_cntl_2 & 7) == 4))
+                else if ((xga->sprite_pos >= 1) || (((xga->disp_cntl_2 & 7) == 2) || (xga->disp_cntl_2 & 7) == 4))
                     xga->cursor_data_on = 1;
                 else if (xga->aperture_cntl == 0) {
                     if (xga->linear_endian_reverse && !(xga->access_mode & 8))
@@ -888,7 +889,7 @@ xga_ext_inb(uint16_t addr, void *priv)
                 d = MIN(s, d);                                                         \
                 break;                                                                 \
             case 0x12:                                                                 \
-                d = MIN(0xff, s + d);                                                  \
+                d = MIN(~0, s + d);                                                  \
                 break;                                                                 \
             case 0x13:                                                                 \
                 d = MAX(0, d - s);                                                     \
@@ -908,7 +909,7 @@ xga_accel_read_pattern_map_pixel(svga_t *svga, int x, int y, int map, uint32_t b
     const xga_t *xga  = &svga->xga;
     uint32_t     addr = base;
     int          bits;
-    uint32_t     byte;
+    uint8_t      byte;
     uint8_t      px;
     int          skip = 0;
 
@@ -978,6 +979,15 @@ xga_accel_read_map_pixel(svga_t *svga, int x, int y, int map, uint32_t base, int
             }
             px = (byte >> bits) & 1;
             return px;
+        case 2: /*4-bit*/
+            addr += (y * (width >> 1));
+            addr += (x >> 1);
+            if (!skip) {
+                READ(addr, byte);
+            } else {
+                byte = mem_readb_phys(addr);
+            }
+            return byte;
         case 3: /*8-bit*/
             addr += (y * width);
             addr += x;
@@ -1060,6 +1070,29 @@ xga_accel_write_map_pixel(svga_t *svga, int x, int y, int map, uint32_t base, ui
             }
             mem_writeb_phys(addr, byte);
             break;
+        case 2: /*4-bit*/
+            addr += (y * (width >> 1));
+            addr += (x >> 1);
+            if (!skip) {
+                READ(addr, byte);
+            } else {
+                byte = mem_readb_phys(addr);
+            }
+            if (xga->linear_endian_reverse) {
+                mask = 0x0f << ((1 - (x & 1)) << 2);
+            } else {
+                if ((xga->accel.px_map_format[map] & 8) && !(xga->access_mode & 8)) {
+                    mask = 0x0f << ((x & 1) << 2);
+                } else {
+                    mask = 0x0f << ((1 - (x & 1)) << 2);
+                }
+            }
+            byte = (byte & ~mask) | (pixel & mask);
+            if (!skip) {
+                WRITE(addr, byte);
+            }
+            mem_writeb_phys(addr, byte);
+            break;
         case 3: /*8-bit*/
             addr += (y * width);
             addr += x;
@@ -1076,14 +1109,12 @@ xga_accel_write_map_pixel(svga_t *svga, int x, int y, int map, uint32_t base, ui
                     pixel = ((pixel & 0xff00) >> 8) | ((pixel & 0x00ff) << 8);
                 else if (xga->access_mode & 8)
                     pixel = ((pixel & 0xff00) >> 8) | ((pixel & 0x00ff) << 8);
-
-                mem_writew_phys(addr, pixel);
             } else {
                 if (!skip) {
                     WRITEW(addr, pixel);
                 }
-                mem_writew_phys(addr, pixel);
             }
+            mem_writew_phys(addr, pixel);
             break;
 
         default:
@@ -1257,7 +1288,7 @@ xga_line_draw_write(svga_t *svga)
 
     err = (xga->accel.bres_err_term);
     if (xga->accel.bres_err_term & 0x2000)
-        destxtmp |= ~0x1fff;
+        err |= ~0x1fff;
 
     if (xga->accel.octant & 0x02) {
         ydir = -1;
@@ -1279,6 +1310,9 @@ xga_line_draw_write(svga_t *svga)
     if (xga->accel.dst_map_y >= 0x1800)
         dy |= ~0x17ff;
 
+    if ((xga->accel.command & 0x30) == 0x30)
+        xga_log("Line Draw Write: BLTWIDTH=%d, BLTHEIGHT=%d, FRGDCOLOR=%04x, XDIR=%i, YDIR=%i, steep=%s, ERR=%04x.\n", xga->accel.blt_width, xga->accel.blt_height, xga->accel.frgd_color & 0xffff, xdir, ydir, (xga->accel.octant & 0x01) ? "0" : "1", err);
+
     if (xga->accel.octant & 0x01) {
         steep = 0;
         SWAP(dx, dy);
@@ -1287,6 +1321,7 @@ xga_line_draw_write(svga_t *svga)
 
     if (xga->accel.pat_src == 8) {
         while (y >= 0) {
+            draw_pixel = 1;
             if (xga->accel.command & 0xc0) {
                 if (steep) {
                     if ((dx >= xga->accel.mask_map_origin_x_off) && (dx <= ((xga->accel.px_map_width[0] & 0xfff) + xga->accel.mask_map_origin_x_off)) && (dy >= xga->accel.mask_map_origin_y_off) && (dy <= ((xga->accel.px_map_height[0] & 0xfff) + xga->accel.mask_map_origin_y_off))) {
@@ -1303,6 +1338,21 @@ xga_line_draw_write(svga_t *svga)
                                 xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
                             else if (((xga->accel.command & 0x30) == 0x20) && y)
                                 xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                            else if ((xga->accel.command & 0x30) == 0x30) {
+                                if (err < 0)
+                                    draw_pixel = 0;
+                                else {
+                                    if (ydir == -1) { /*Bottom-to-Top*/
+                                        if (!x)
+                                            draw_pixel = 0;
+                                    } else { /*Top-to-Bottom*/
+                                        if (!y)
+                                            draw_pixel = 0;
+                                    }
+                                }
+                                if (draw_pixel)
+                                    xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                            }
                         }
                     }
                 } else {
@@ -1320,6 +1370,17 @@ xga_line_draw_write(svga_t *svga)
                                 xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
                             else if (((xga->accel.command & 0x30) == 0x20) && y)
                                 xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                            else if ((xga->accel.command & 0x30) == 0x30) {
+                                if (xdir == -1) { /*Bottom-to-Top*/
+                                    if (!x)
+                                        draw_pixel = 0;
+                                } else { /*Top-to-Bottom*/
+                                    if (!y)
+                                        draw_pixel = 0;
+                                }
+                                if (draw_pixel)
+                                    xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                            }
                         }
                     }
                 }
@@ -1338,6 +1399,21 @@ xga_line_draw_write(svga_t *svga)
                             xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
                         else if (((xga->accel.command & 0x30) == 0x20) && y)
                             xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                        else if ((xga->accel.command & 0x30) == 0x30) {
+                            if (err < 0)
+                                draw_pixel = 0;
+                            else {
+                                if (ydir == -1) { /*Bottom-to-Top*/
+                                    if (!x)
+                                        draw_pixel = 0;
+                                } else { /*Top-to-Bottom*/
+                                    if (!y)
+                                        draw_pixel = 0;
+                                }
+                            }
+                            if (draw_pixel)
+                                xga_accel_write_map_pixel(svga, dx, dy, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                        }
                     }
                 } else {
                     src_dat  = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.src_map_x & 0xfff, xga->accel.src_map_y & 0xfff, xga->accel.src_map, srcbase, xga->accel.px_map_width[xga->accel.src_map] + 1, 1) : xga->accel.frgd_color;
@@ -1353,6 +1429,17 @@ xga_line_draw_write(svga_t *svga)
                             xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
                         else if (((xga->accel.command & 0x30) == 0x20) && y)
                             xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                        else if ((xga->accel.command & 0x30) == 0x30) {
+                            if (xdir == -1) { /*Bottom-to-Top*/
+                                if (!x)
+                                    draw_pixel = 0;
+                            } else { /*Top-to-Bottom*/
+                                if (!y)
+                                    draw_pixel = 0;
+                            }
+                            if (draw_pixel)
+                                xga_accel_write_map_pixel(svga, dy, dx, xga->accel.dst_map, dstbase, dest_dat, xga->accel.px_map_width[xga->accel.dst_map] + 1);
+                        }
                     }
                 }
             }
@@ -1361,7 +1448,7 @@ xga_line_draw_write(svga_t *svga)
                 break;
             }
 
-            while (err > 0) {
+            while (err >= 0) {
                 dy += ydir;
                 err -= (dmajor << 1);
             }
@@ -1383,10 +1470,13 @@ xga_line_draw_write(svga_t *svga)
     }
 }
 
+#undef SWAP
+
 static void
 xga_bitblt(svga_t *svga)
 {
     xga_t   *xga = &svga->xga;
+    uint8_t  area_state = 0;
     uint32_t src_dat;
     uint32_t dest_dat;
     uint32_t old_dest_dat;
@@ -1405,9 +1495,12 @@ xga_bitblt(svga_t *svga)
 #endif
     uint32_t frgdcol = xga->accel.frgd_color;
     uint32_t bkgdcol = xga->accel.bkgd_color;
-    int      mix     = 0;
+    int      mix  = 0;
     int      xdir;
     int      ydir;
+    int      skip = 0;
+    int      x = 0;
+    int      y = 0;
 
     if (xga->accel.octant & 0x02) {
         ydir = -1;
@@ -1556,17 +1649,105 @@ xga_bitblt(svga_t *svga)
         xga_log("Pattern Map = %d: CMD = %08x: PATBase = %08x, SRCBase = %08x, DSTBase = %08x\n",
                 xga->accel.pat_src, xga->accel.command, patbase, srcbase, dstbase);
         xga_log("CMD = %08x: Y = %d, X = %d, patsrc = %02x, srcmap = %d, dstmap = %d, py = %d, "
-                "sy = %d, dy = %d, width0 = %d, width1 = %d, width2 = %d, width3 = %d\n",
+                "sy = %d, dy = %d, width0 = %d, width1 = %d, width2 = %d, width3 = %d, bkgdcol = %02x\n",
                 xga->accel.command, xga->accel.y, xga->accel.x, xga->accel.pat_src,
                 xga->accel.src_map, xga->accel.dst_map, xga->accel.py, xga->accel.sy, xga->accel.dy,
                 xga->accel.px_map_width[0], xga->accel.px_map_width[1],
-                xga->accel.px_map_width[2], xga->accel.px_map_width[3]);
+                xga->accel.px_map_width[2], xga->accel.px_map_width[3], bkgdcol);
 
-        while (xga->accel.y >= 0) {
-            mix = xga_accel_read_pattern_map_pixel(svga, xga->accel.px, xga->accel.py, xga->accel.pat_src, patbase, patwidth + 1);
+        if (((xga->accel.command >> 24) & 0x0f) == 0x0a) {
+            while (xga->accel.y >= 0) {
+                mix = xga_accel_read_pattern_map_pixel(svga, xga->accel.px, xga->accel.py, xga->accel.pat_src, patbase, patwidth + 1);
+                if (mix)
+                    area_state ^= 1;
 
-            if (xga->accel.command & 0xc0) {
-                if ((xga->accel.dx >= xga->accel.mask_map_origin_x_off) && (xga->accel.dx <= ((xga->accel.px_map_width[0] & 0xfff) + xga->accel.mask_map_origin_x_off)) && (xga->accel.dy >= xga->accel.mask_map_origin_y_off) && (xga->accel.dy <= ((xga->accel.px_map_height[0] & 0xfff) + xga->accel.mask_map_origin_y_off))) {
+                if (xga->accel.command & 0xc0) {
+                    if ((xga->accel.dx >= xga->accel.mask_map_origin_x_off) && (xga->accel.dx <= ((xga->accel.px_map_width[0] & 0xfff) + xga->accel.mask_map_origin_x_off)) && (xga->accel.dy >= xga->accel.mask_map_origin_y_off) && (xga->accel.dy <= ((xga->accel.px_map_height[0] & 0xfff) + xga->accel.mask_map_origin_y_off))) {
+                        if (area_state)
+                            src_dat = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : frgdcol;
+                        else
+                            src_dat = (((xga->accel.command >> 30) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : bkgdcol;
+
+                        if ((xga->accel.px_map_format[xga->accel.dst_map] & 7) <= 3)
+                            src_dat &= 0xff;
+
+                        dest_dat = xga_accel_read_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dstwidth + 1, 0);
+                        if ((xga->accel.cc_cond == 4) || ((xga->accel.cc_cond == 1) && (dest_dat > color_cmp)) || ((xga->accel.cc_cond == 2) && (dest_dat == color_cmp)) || ((xga->accel.cc_cond == 3) && (dest_dat < color_cmp)) || ((xga->accel.cc_cond == 5) && (dest_dat >= color_cmp)) || ((xga->accel.cc_cond == 6) && (dest_dat != color_cmp)) || ((xga->accel.cc_cond == 7) && (dest_dat <= color_cmp))) {
+                            old_dest_dat = dest_dat;
+                            ROP(area_state, dest_dat, src_dat);
+                            dest_dat = (dest_dat & plane_mask) | (old_dest_dat & ~plane_mask);
+                            xga_log("1SRCDat=%02x, DSTDat=%02x, Old=%02x, MIX=%d.\n", src_dat, dest_dat, old_dest_dat, area_state);
+                            xga_accel_write_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dest_dat, dstwidth + 1);
+                        }
+                    }
+                } else {
+                    if (area_state)
+                        src_dat = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : frgdcol;
+                    else
+                        src_dat = (((xga->accel.command >> 30) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : bkgdcol;
+
+                    if ((xga->accel.px_map_format[xga->accel.dst_map] & 7) <= 3)
+                        src_dat &= 0xff;
+
+                    dest_dat = xga_accel_read_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dstwidth + 1, 0);
+                    if ((xga->accel.cc_cond == 4) || ((xga->accel.cc_cond == 1) && (dest_dat > color_cmp)) || ((xga->accel.cc_cond == 2) && (dest_dat == color_cmp)) || ((xga->accel.cc_cond == 3) && (dest_dat < color_cmp)) || ((xga->accel.cc_cond == 5) && (dest_dat >= color_cmp)) || ((xga->accel.cc_cond == 6) && (dest_dat != color_cmp)) || ((xga->accel.cc_cond == 7) && (dest_dat <= color_cmp))) {
+                        old_dest_dat = dest_dat;
+                        ROP(area_state, dest_dat, src_dat);
+                        dest_dat = (dest_dat & plane_mask) | (old_dest_dat & ~plane_mask);
+                        xga_log("2Fill: NumXY(%d,%d): DXY(%d,%d): SRCDat=%02x, DSTDat=%02x, Old=%02x, frgdcol=%02x, bkgdcol=%02x, MIX=%d, frgdmix=%02x, bkgdmix=%02x, dstmapfmt=%02x, srcmapfmt=%02x, srcmapnum=%d.\n", x, y, xga->accel.dx, xga->accel.dy, src_dat, dest_dat, old_dest_dat, frgdcol, bkgdcol, area_state, xga->accel.frgd_mix & 0x1f, xga->accel.bkgd_mix & 0x1f, xga->accel.px_map_format[xga->accel.dst_map] & 0x0f, xga->accel.px_map_format[xga->accel.src_map] & 0x0f, xga->accel.src_map);
+                        xga_accel_write_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dest_dat, dstwidth + 1);
+                    }
+                }
+
+                xga->accel.sx = ((xga->accel.sx + 1) & srcwidth) | (xga->accel.sx & ~srcwidth);
+                xga->accel.px = ((xga->accel.px + 1) & patwidth) | (xga->accel.px & ~patwidth);
+                xga->accel.dx++;
+                xga->accel.x--;
+                x++;
+                if (xga->accel.x < 0) {
+                    area_state = 0;
+                    x = 0;
+                    xga->accel.y--;
+                    xga->accel.x = xga->accel.blt_width & 0xfff;
+
+                    xga->accel.dx = xga->accel.dst_map_x & 0x1fff;
+                    if (xga->accel.dst_map_x >= 0x1800)
+                        xga->accel.dx |= ~0x17ff;
+                    xga->accel.sx = xga->accel.src_map_x & 0xfff;
+                    xga->accel.px = xga->accel.pat_map_x & 0xfff;
+
+                    xga->accel.sy = ((xga->accel.sy + ydir) & srcheight) | (xga->accel.sy & ~srcheight);
+                    xga->accel.py += ydir;
+                    xga->accel.dy += ydir;
+                    y++;
+
+                    if (xga->accel.y < 0) {
+                        xga->accel.dst_map_x = xga->accel.dx;
+                        xga->accel.dst_map_y = xga->accel.dy;
+                        return;
+                    }
+                }
+            }
+        } else {
+            while (xga->accel.y >= 0) {
+                mix = xga_accel_read_pattern_map_pixel(svga, xga->accel.px, xga->accel.py, xga->accel.pat_src, patbase, patwidth + 1);
+
+                if (xga->accel.command & 0xc0) {
+                    if ((xga->accel.dx >= xga->accel.mask_map_origin_x_off) && (xga->accel.dx <= ((xga->accel.px_map_width[0] & 0xfff) + xga->accel.mask_map_origin_x_off)) && (xga->accel.dy >= xga->accel.mask_map_origin_y_off) && (xga->accel.dy <= ((xga->accel.px_map_height[0] & 0xfff) + xga->accel.mask_map_origin_y_off))) {
+                        if (mix) {
+                            src_dat = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : frgdcol;
+                        } else {
+                            src_dat = (((xga->accel.command >> 30) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : bkgdcol;
+                        }
+                        dest_dat = xga_accel_read_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dstwidth + 1, 0);
+                        if ((xga->accel.cc_cond == 4) || ((xga->accel.cc_cond == 1) && (dest_dat > color_cmp)) || ((xga->accel.cc_cond == 2) && (dest_dat == color_cmp)) || ((xga->accel.cc_cond == 3) && (dest_dat < color_cmp)) || ((xga->accel.cc_cond == 5) && (dest_dat >= color_cmp)) || ((xga->accel.cc_cond == 6) && (dest_dat != color_cmp)) || ((xga->accel.cc_cond == 7) && (dest_dat <= color_cmp))) {
+                            old_dest_dat = dest_dat;
+                            ROP(mix, dest_dat, src_dat);
+                            dest_dat = (dest_dat & plane_mask) | (old_dest_dat & ~plane_mask);
+                            xga_accel_write_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dest_dat, dstwidth + 1);
+                        }
+                    }
+                } else {
                     if (mix) {
                         src_dat = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : frgdcol;
                     } else {
@@ -1580,49 +1761,36 @@ xga_bitblt(svga_t *svga)
                         xga_accel_write_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dest_dat, dstwidth + 1);
                     }
                 }
-            } else {
-                if (mix) {
-                    src_dat = (((xga->accel.command >> 28) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : frgdcol;
-                } else {
-                    src_dat = (((xga->accel.command >> 30) & 3) == 2) ? xga_accel_read_map_pixel(svga, xga->accel.sx, xga->accel.sy, xga->accel.src_map, srcbase, srcwidth + 1, 1) : bkgdcol;
-                }
-                dest_dat = xga_accel_read_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dstwidth + 1, 0);
-                if ((xga->accel.cc_cond == 4) || ((xga->accel.cc_cond == 1) && (dest_dat > color_cmp)) || ((xga->accel.cc_cond == 2) && (dest_dat == color_cmp)) || ((xga->accel.cc_cond == 3) && (dest_dat < color_cmp)) || ((xga->accel.cc_cond == 5) && (dest_dat >= color_cmp)) || ((xga->accel.cc_cond == 6) && (dest_dat != color_cmp)) || ((xga->accel.cc_cond == 7) && (dest_dat <= color_cmp))) {
-                    old_dest_dat = dest_dat;
-                    ROP(mix, dest_dat, src_dat);
-                    dest_dat = (dest_dat & plane_mask) | (old_dest_dat & ~plane_mask);
-                    xga_accel_write_map_pixel(svga, xga->accel.dx, xga->accel.dy, xga->accel.dst_map, dstbase, dest_dat, dstwidth + 1);
-                }
-            }
 
-            xga->accel.sx += xdir;
-            if (xga->accel.pattern)
-                xga->accel.px = ((xga->accel.px + xdir) & patwidth) | (xga->accel.px & ~patwidth);
-            else
-                xga->accel.px += xdir;
-            xga->accel.dx += xdir;
-            xga->accel.x--;
-            if (xga->accel.x < 0) {
-                xga->accel.y--;
-                xga->accel.x = (xga->accel.blt_width & 0xfff);
-
-                xga->accel.dx = xga->accel.dst_map_x & 0x1fff;
-                if (xga->accel.dst_map_x >= 0x1800)
-                    xga->accel.dx |= ~0x17ff;
-                xga->accel.sx = xga->accel.src_map_x & 0xfff;
-                xga->accel.px = xga->accel.pat_map_x & 0xfff;
-
-                xga->accel.sy += ydir;
+                xga->accel.sx += xdir;
                 if (xga->accel.pattern)
-                    xga->accel.py = ((xga->accel.py + ydir) & patheight) | (xga->accel.py & ~patheight);
+                    xga->accel.px = ((xga->accel.px + xdir) & patwidth) | (xga->accel.px & ~patwidth);
                 else
-                    xga->accel.py += ydir;
-                xga->accel.dy += ydir;
+                    xga->accel.px += xdir;
+                xga->accel.dx += xdir;
+                xga->accel.x--;
+                if (xga->accel.x < 0) {
+                    xga->accel.y--;
+                    xga->accel.x = (xga->accel.blt_width & 0xfff);
 
-                if (xga->accel.y < 0) {
-                    xga->accel.dst_map_x = xga->accel.dx;
-                    xga->accel.dst_map_y = xga->accel.dy;
-                    return;
+                    xga->accel.dx = xga->accel.dst_map_x & 0x1fff;
+                    if (xga->accel.dst_map_x >= 0x1800)
+                        xga->accel.dx |= ~0x17ff;
+                    xga->accel.sx = xga->accel.src_map_x & 0xfff;
+                    xga->accel.px = xga->accel.pat_map_x & 0xfff;
+
+                    xga->accel.sy += ydir;
+                    if (xga->accel.pattern)
+                        xga->accel.py = ((xga->accel.py + ydir) & patheight) | (xga->accel.py & ~patheight);
+                    else
+                        xga->accel.py += ydir;
+                    xga->accel.dy += ydir;
+
+                    if (xga->accel.y < 0) {
+                        xga->accel.dst_map_x = xga->accel.dx;
+                        xga->accel.dst_map_y = xga->accel.dy;
+                        return;
+                    }
                 }
             }
         }
@@ -2054,24 +2222,29 @@ exec_command:
 #endif
 
                     switch ((xga->accel.command >> 24) & 0x0f) {
+                        case 2: /*Short Stroke Vectors Read */
+                            xga_log("Short Stroke Vectors Read.\n");
+                            break;
                         case 3: /*Bresenham Line Draw Read*/
                             xga_log("Line Draw Read\n");
                             break;
-                        case 4: /*Short Stroke Vectors*/
-                            xga_log("Short Stroke Vectors.\n");
+                        case 4: /*Short Stroke Vectors Write*/
+                            xga_log("Short Stroke Vectors Write.\n");
                             break;
                         case 5: /*Bresenham Line Draw Write*/
                             xga_log("Line Draw Write.\n");
                             xga_line_draw_write(svga);
                             break;
                         case 8: /*BitBLT*/
+                            xga_log("BitBLT.\n");
                             xga_bitblt(svga);
                             break;
                         case 9: /*Inverting BitBLT*/
                             xga_log("Inverting BitBLT\n");
                             break;
                         case 0x0a: /*Area Fill*/
-                            xga_log("Area Fill.\n");
+                            xga_log("Area Fill BitBLT.\n");
+                            xga_bitblt(svga);
                             break;
 
                         default:
@@ -2355,6 +2528,47 @@ xga_render_overscan_right(xga_t *xga, svga_t *svga)
         *line_ptr++ = svga->overscan_color;
 }
 
+static void
+xga_render_4bpp(svga_t *svga)
+{
+    xga_t *xga = &svga->xga;
+    uint32_t *p;
+    uint32_t  dat;
+
+    if ((xga->displine + svga->y_add) < 0)
+        return;
+
+    if (xga->changedvram[xga->ma >> 12] || xga->changedvram[(xga->ma >> 12) + 1] || svga->fullchange) {
+        p = &svga->monitor->target_buffer->line[xga->displine + svga->y_add][svga->x_add];
+
+        if (xga->firstline_draw == 2000)
+            xga->firstline_draw = xga->displine;
+        xga->lastline_draw = xga->displine;
+
+        for (int x = 0; x <= xga->h_disp; x += 16) {
+            dat  = *(uint32_t *) (&xga->vram[xga->ma & xga->vram_mask]);
+            p[0] = xga->pallook[(dat >> 4) & 0x0f];
+            p[1] = xga->pallook[dat & 0x0f];
+            p[2] = xga->pallook[(dat >> 12) & 0x0f];
+            p[3] = xga->pallook[(dat >> 8) & 0x0f];
+            p[4] = xga->pallook[(dat >> 20) & 0x0f];
+            p[5] = xga->pallook[(dat >> 16) & 0x0f];
+            p[6] = xga->pallook[(dat >> 28) & 0x0f];
+            p[7] = xga->pallook[(dat >> 24) & 0x0f];
+
+            dat  = *(uint32_t *) (&xga->vram[(xga->ma + 4) & xga->vram_mask]);
+            p[9] = xga->pallook[dat & 0x0f];
+            p[11] = xga->pallook[(dat >> 8) & 0x0f];
+            p[13] = xga->pallook[(dat >> 16) & 0x0f];
+            p[15] = xga->pallook[(dat >> 24) & 0x0f];
+
+            xga->ma += 8;
+            p += 16;
+        }
+        xga->ma &= xga->vram_mask;
+    }
+}
+
 static void
 xga_render_8bpp(svga_t *svga)
 {
@@ -2449,7 +2663,7 @@ xga_write(uint32_t addr, uint8_t val, void *priv)
     if (addr >= xga->vram_size)
         return;
 
-    cycles -= video_timing_write_b;
+    cycles -= svga->monitor->mon_video_timing_write_b;
 
     if (xga->access_mode & 8) {
         if ((xga->access_mode & 7) == 4)
@@ -2510,7 +2724,7 @@ xga_read(uint32_t addr, void *priv)
     if (addr >= xga->vram_size)
         return ret;
 
-    cycles -= video_timing_read_b;
+    cycles -= svga->monitor->mon_video_timing_read_b;
 
     if (xga->access_mode & 8) {
         if ((xga->access_mode & 7) == 4)
@@ -2695,6 +2909,9 @@ xga_do_render(svga_t *svga)
 
     xga_log("DISPCNTL = %d, vga = %d.\n", xga->disp_cntl_2 & 7, vga_on);
     switch (xga->disp_cntl_2 & 7) {
+        case 2:
+            xga_render_4bpp(svga);
+            break;
         case 3:
             xga_render_8bpp(svga);
             break;
@@ -2773,15 +2990,9 @@ xga_poll(xga_t *xga, svga_t *svga)
             if (xga->sc == xga->rowcount) {
                 xga->sc = 0;
 
-                if ((xga->disp_cntl_2 & 7) == 4) {
-                    xga->maback += (xga->rowoffset << 4);
-                    if (xga->interlace)
-                        xga->maback += (xga->rowoffset << 4);
-                } else {
-                    xga->maback += (xga->rowoffset << 3);
-                    if (xga->interlace)
-                        xga->maback += (xga->rowoffset << 3);
-                }
+                xga->maback += (xga->rowoffset << (xga->disp_cntl_2 & 7));
+                if (xga->interlace)
+                    xga->maback += (xga->rowoffset << (xga->disp_cntl_2 & 7));
                 xga->maback &= xga->vram_mask;
                 xga->ma = xga->maback;
             } else {