summaryrefslogtreecommitdiff
path: root/shared-core/savage_bci.c
AgeCommit message (Collapse)Author
2006-12-19make a savage function static from kernelDave Airlie
2006-10-16dev->agp_buffer_map is not initialized for AGP DMA on savagesMichael Karcher
bug 8662
2006-04-09Revert a change that accidentally went in with whitespace changes fromEric Anholt
Linux, which broke on FreeBSD. DRM_COPY_*_IOCTL checks for the size parameter matching the ioctl's command size there, since the copin/out happened earlier.
2006-01-02whitespace cleanup/aligment with kernelDave Airlie
2005-11-08Initial port of savage to FreeBSD for the AGP and !ShadowStatus case. AddsEric Anholt
drm_mtrr_{add,del} for handling the MTRR setup. Still has a LOR issue with DRM_VERIFYAREA_READ/DRM_COPY_FROM_USER_UNCHECKED in savage_bci.c -- this won't work with the fine-grained locking in use, and just doing a single copyin to a temporary will probably work fine. Also note that the module leaks approximately 4 kb on unload.
2005-09-03convert ioctl flags to use flags instead of separate intsDave Airlie
2005-08-09Make sure savage has 3rd ioctl parameterJon Smirl
2005-08-07make some functions static in the savage drm driverDave Airlie
2005-08-05Rename the driver hooks in the DRM to something a little moreEric Anholt
understandable: preinit -> load postinit -> (removed) presetup -> firstopen postsetup -> (removed) open_helper -> open prerelease -> preclose free_filp_priv -> postclose pretakedown -> lastclose postcleanup -> unload release -> reclaim_buffers_locked version -> (removed) postinit and version were replaced with generic code in the Linux DRM (drivers now set their version numbers and description in the driver structure, like on BSD). postsetup wasn't used at all. Fixes the savage hooks for initializing and tearing down mappings at the right times. Testing involved at least starting X, running glxgears, killing glxgears, exiting X, and repeating. Tested on: FreeBSD (g200, g400, r200, r128) Linux (r200, savage4)
2005-06-28- Remove drm_initmap and replace its usage with drm_addmap. This reducesEric Anholt
code duplication, and it also hands you the map pointer so you don't need to re-find it. - Remove the permanent maps flag. Instead, for register and framebuffer maps, we always check whether there's already a map of that type and offset around. Move the Radeon map initialization into presetup (first open) so it happens again after every takedown. - Remove the split cleanup of maps between driver takedown (last close) and cleanup (module unload). Instead, always tear down maps on takedown, and drivers can recreate them on first open. - Make MGA always use addmap, instead of allocating consistent memory in the PCI case and then faking up a map for it, which accomplished nearly the same thing, in a different order. Note that the maps are exposed to the user again: we may want to expose a flag to avoid this, but it's not a security concern, and saves us a lot of code. - Remove rmmaps in the MGA driver. Since the function is only called during takedown anyway, we can let them die a natural death. - Make removal of maps happen in one function, which is called by both drm_takedown and drm_rmmap_ioctl. Reviewed by: idr (previous revision) Tested on: mga (old/new/pci dma), radeon, savage
2005-06-21Change initialization of savage register access to _DRM_READ_ONLY. Flags ofJon Smirl
zero does not mean no privs, instead it grants write access _DRM_READ_ONLY only applies to non-root users. Problem is only in CVS, initmaps are not in the kernel yet.
2005-06-04Remove warnings about code mixed with declerations..Dave Airlie
2005-03-13Command DMA optimizations:Felix Kuehling
- don't waste DMA memory when small command buffers are flushed - minimized padding with noops - slightly simplified faked DMA flushing
2005-03-07Tracked down random lockups related to command DMA that occurred in Quake3Felix Kuehling
after ~5min: buffer aging was subtly broken. Part of this may have also affected vertex DMA buffer aging and client-side texture heap aging, though with less fatal consequences. Bumped minor version and driver date.
2005-03-06Added support for command DMA on Savage4-based hardware. UnfortunatelyFelix Kuehling
command and vertex DMA don't work at the same time. Command DMA performance is superior and works with all vertex formats. Bumped minor version and driver date.
2005-02-23Use wrap counter to extend 16-bit hardware event tags to 32-bit logicalFelix Kuehling
event counters in the SAVAGE_EVENT_EMIT/WAIT ioctls. This is needed for reliable client-side texture heap aging. Slightly simplified DMA buffer aging while at it. Bumped minor version and driver date.
2005-02-05- Implement drm_initmap, and extend it with the resource number to helpEric Anholt
FreeBSD. Add drm_get_resource_{start|len} so linux-specific stuff doesn't need to be in shared code. - Fix mach64 build by using __DECONST to work around passing a const pointer to useracc, which is unfortunately not marked const. - Get rid of a lot of maplist code by not having dev->maplist be a pointer, and by sticking the link entries directly in drm_local_map_t rather than having a separate structure for the linked list. - Factor out map uninit and removal into its own routine, rather than duplicating in both drm_takedown() and drm_rmmap(). - Hook up more driver functions, and correct FreeBSD-specific bits of radeon_cp.c, making radeon work. - Baby steps towards using bus_space as we should.
2005-01-15Setup MTRRs for frame buffer and aperture manually on Savage3D andFelix Kuehling
Savage4-based cards. Automatic setup in drm_initmap doesn't work due to the weird alignment and size of the aperture.
2005-01-10Only try to find the agp_buffer_map if dma_type is AGP. This is all that'sFelix Kuehling
needed on the DRM side to support PCI Savages. Bumped patch level and driver date.
2005-01-08Fixed off-by-one error in savage_bci_wait_fifo_shadow.Felix Kuehling
2005-01-053D scissor regs are now managed by the DRM to iterate over clip rectsFelix Kuehling
passed to the cmdbuf ioctl (try xeyes on top of glxgears ;-) Tightened the texture state check Bumped Savage DRM version to 2.1.0 so that DRI driver can (theoretically) depend on the DRM to manage the scissor registers
2005-01-01Completeley rewritten Savage DRM which can be considered secure (moduloFelix Kuehling
implementation errors). Direct hardware (MMIO, BCI) access is no longer needed in the Mesa driver. Bumped version to 2.0.0. Corresponding changes to the DDX and Mesa drivers are being committed.
7' href='#n417'>417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
/*
 * Copyright 2004 The Unichrome Project. All Rights Reserved.
 * Copyright 2005 Thomas Hellstrom. All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sub license,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHOR(S), AND/OR THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 *
 * Author: Thomas Hellstrom 2004, 2005.
 * This code was written using docs obtained under NDA from VIA Inc.
 *
 * Don't run this code directly on an AGP buffer. Due to cache problems it will
 * be very slow.
 */

#include "via_3d_reg.h"
#include "drmP.h"
#include "drm.h"
#include "via_drm.h"
#include "via_verifier.h"
#include "via_drv.h"

typedef enum {
	state_command,
	state_header2,
	state_header1,
	state_vheader5,
	state_vheader6,
	state_error
} verifier_state_t;

typedef enum {
	no_check = 0,
	check_for_header2,
	check_for_header1,
	check_for_header2_err,
	check_for_header1_err,
	check_for_fire,
	check_z_buffer_addr0,
	check_z_buffer_addr1,
	check_z_buffer_addr_mode,
	check_destination_addr0,
	check_destination_addr1,
	check_destination_addr_mode,
	check_for_dummy,
	check_for_dd,
	check_texture_addr0,
	check_texture_addr1,
	check_texture_addr2,
	check_texture_addr3,
	check_texture_addr4,
	check_texture_addr5,
	check_texture_addr6,
	check_texture_addr7,
	check_texture_addr8,
	check_texture_addr_mode,
	check_for_vertex_count,
	check_number_texunits,
	forbidden_command
} hazard_t;

/*
 * Associates each hazard above with a possible multi-command
 * sequence. For example an address that is split over multiple
 * commands and that needs to be checked at the first command 
 * that does not include any part of the address.
 */

static drm_via_sequence_t seqs[] = {
	no_sequence,
	no_sequence,
	no_sequence,
	no_sequence,
	no_sequence,
	no_sequence,
	z_address,
	z_address,
	z_address,
	dest_address,
	dest_address,
	dest_address,
	no_sequence,
	no_sequence,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	tex_address,
	no_sequence
};

typedef struct {
	unsigned int code;
	hazard_t hz;
} hz_init_t;

static hz_init_t init_table1[] = {
	{0xf2, check_for_header2_err},
	{0xf0, check_for_header1_err},
	{0xee, check_for_fire},
	{0xcc, check_for_dummy},
	{0xdd, check_for_dd},
	{0x00, no_check},
	{0x10, check_z_buffer_addr0},
	{0x11, check_z_buffer_addr1},
	{0x12, check_z_buffer_addr_mode},
	{0x13, no_check},
	{0x14, no_check},
	{0x15, no_check},
	{0x23, no_check},
	{0x24, no_check},
	{0x33, no_check},
	{0x34, no_check},
	{0x35, no_check},
	{0x36, no_check},
	{0x37, no_check},
	{0x38, no_check},
	{0x39, no_check},
	{0x3A, no_check},
	{0x3B, no_check},
	{0x3C, no_check},
	{0x3D, no_check},
	{0x3E, no_check},
	{0x40, check_destination_addr0},
	{0x41, check_destination_addr1},
	{0x42, check_destination_addr_mode},
	{0x43, no_check},
	{0x44, no_check},
	{0x50, no_check},
	{0x51, no_check},
	{0x52, no_check},
	{0x53, no_check},
	{0x54, no_check},
	{0x55, no_check},
	{0x56, no_check},
	{0x57, no_check},
	{0x58, no_check},
	{0x70, no_check},
	{0x71, no_check},
	{0x78, no_check},
	{0x79, no_check},
	{0x7A, no_check},
	{0x7B, no_check},
	{0x7C, no_check},
	{0x7D, check_for_vertex_count}
};

static hz_init_t init_table2[] = {
	{0xf2, check_for_header2_err},
	{0xf0, check_for_header1_err},
	{0xee, check_for_fire},
	{0xcc, check_for_dummy},
	{0x00, check_texture_addr0},
	{0x01, check_texture_addr0},
	{0x02, check_texture_addr0},
	{0x03, check_texture_addr0},
	{0x04, check_texture_addr0},
	{0x05, check_texture_addr0},
	{0x06, check_texture_addr0},
	{0x07, check_texture_addr0},
	{0x08, check_texture_addr0},
	{0x09, check_texture_addr0},
	{0x20, check_texture_addr1},
	{0x21, check_texture_addr1},
	{0x22, check_texture_addr1},
	{0x23, check_texture_addr4},
	{0x2B, check_texture_addr3},
	{0x2C, check_texture_addr3},
	{0x2D, check_texture_addr3},
	{0x2E, check_texture_addr3},
	{0x2F, check_texture_addr3},
	{0x30, check_texture_addr3},
	{0x31, check_texture_addr3},
	{0x32, check_texture_addr3},
	{0x33, check_texture_addr3},
	{0x34, check_texture_addr3},
	{0x4B, check_texture_addr5},
	{0x4C, check_texture_addr6},
	{0x51, check_texture_addr7},
	{0x52, check_texture_addr8},
	{0x77, check_texture_addr2},
	{0x78, no_check},
	{0x79, no_check},
	{0x7A, no_check},
	{0x7B, check_texture_addr_mode},
	{0x7C, no_check},
	{0x7D, no_check},
	{0x7E, no_check},
	{0x7F, no_check},
	{0x80, no_check},
	{0x81, no_check},
	{0x82, no_check},
	{0x83, no_check},
	{0x85, no_check},
	{0x86, no_check},
	{0x87, no_check},
	{0x88, no_check},
	{0x89, no_check},
	{0x8A, no_check},
	{0x90, no_check},
	{0x91, no_check},
	{0x92, no_check},
	{0x93, no_check}
};

static hz_init_t init_table3[] = {
	{0xf2, check_for_header2_err},
	{0xf0, check_for_header1_err},
	{0xcc, check_for_dummy},
	{0x00, check_number_texunits}
};

static hazard_t table1[256];
static hazard_t table2[256];
static hazard_t table3[256];

static __inline__ int
eat_words(const uint32_t ** buf, const uint32_t * buf_end, unsigned num_words)
{
	if ((buf_end - *buf) >= num_words) {
		*buf += num_words;
		return 0;
	}
	DRM_ERROR("Illegal termination of DMA command buffer\n");
	return 1;
}

/*
 * Partially stolen from drm_memory.h
 */

static __inline__ drm_local_map_t *via_drm_lookup_agp_map(drm_via_state_t * seq,
							  unsigned long offset,
							  unsigned long size,
							  drm_device_t * dev)
{
#ifdef __linux__
	struct list_head *list;
	drm_map_list_t *r_list;
#endif
	drm_local_map_t *map = seq->map_cache;

	if (map && map->offset <= offset
	    && (offset + size) <= (map->offset + map->size)) {
		return map;
	}
#ifdef __linux__
	list_for_each(list, &dev->maplist->head) {
		r_list = (drm_map_list_t *) list;
		map = r_list->map;
		if (!map)
			continue;
#else
	TAILQ_FOREACH(map, &dev->maplist, link) {
#endif
		if (map->offset <= offset
		    && (offset + size) <= (map->offset + map->size)
		    && !(map->flags & _DRM_RESTRICTED)
		    && (map->type == _DRM_AGP)) {
			seq->map_cache = map;
			return map;
		}
	}
	return NULL;
}

/*
 * Require that all AGP texture levels reside in the same AGP map which should 
 * be mappable by the client. This is not a big restriction.
 * FIXME: To actually enforce this security policy strictly, drm_rmmap 
 * would have to wait for dma quiescent before removing an AGP map. 
 * The via_drm_lookup_agp_map call in reality seems to take
 * very little CPU time.
 */

static __inline__ int finish_current_sequence(drm_via_state_t * cur_seq)
{
	switch (cur_seq->unfinished) {
	case z_address:
		DRM_DEBUG("Z Buffer start address is 0x%x\n", cur_seq->z_addr);
		break;
	case dest_address:
		DRM_DEBUG("Destination start address is 0x%x\n",
			  cur_seq->d_addr);
		break;
	case tex_address:
		if (cur_seq->agp_texture) {
			unsigned start =
			    cur_seq->tex_level_lo[cur_seq->texture];
			unsigned end = cur_seq->tex_level_hi[cur_seq->texture];
			unsigned long lo = ~0, hi = 0, tmp;
			uint32_t *addr, *pitch, *height, tex;
			unsigned i;
			int npot;

			if (end > 9)
				end = 9;
			if (start > 9)
				start = 9;

			addr =
			    &(cur_seq->t_addr[tex = cur_seq->texture][start]);
			pitch = &(cur_seq->pitch[tex][start]);
			height = &(cur_seq->height[tex][start]);
			npot = cur_seq->tex_npot[tex];
			for (i = start; i <= end; ++i) {
				tmp = *addr++;
				if (tmp < lo)
					lo = tmp;
				if (i == 0 && npot)
					tmp += (*height++ * *pitch++);
				else
					tmp += (*height++ << *pitch++);
				if (tmp > hi)
					hi = tmp;
			}

			if (!via_drm_lookup_agp_map
			    (cur_seq, lo, hi - lo, cur_seq->dev)) {
				DRM_ERROR
				    ("AGP texture is not in allowed map\n");
				return 2;
			}
		}
		break;
	default:
		break;
	}
	cur_seq->unfinished = no_sequence;
	return 0;
}

static __inline__ int
investigate_hazard(uint32_t cmd, hazard_t hz, drm_via_state_t * cur_seq)
{
	register uint32_t tmp, *tmp_addr;

	if (cur_seq->unfinished && (cur_seq->unfinished != seqs[hz])) {
		int ret;
		if ((ret = finish_current_sequence(cur_seq)))
			return ret;
	}

	switch (hz) {
	case check_for_header2:
		if (cmd == HALCYON_HEADER2)
			return 1;
		return 0;
	case check_for_header1:
		if ((cmd & HALCYON_HEADER1MASK) == HALCYON_HEADER1)
			return 1;
		return 0;
	case check_for_header2_err:
		if (cmd == HALCYON_HEADER2)
			return 1;
		DRM_ERROR("Illegal DMA HALCYON_HEADER2 command\n");
		break;
	case check_for_header1_err:
		if ((cmd & HALCYON_HEADER1MASK) == HALCYON_HEADER1)
			return 1;
		DRM_ERROR("Illegal DMA HALCYON_HEADER1 command\n");
		break;
	case check_for_fire:
		if ((cmd & HALCYON_FIREMASK) == HALCYON_FIRECMD)
			return 1;
		DRM_ERROR("Illegal DMA HALCYON_FIRECMD command\n");
		break;
	case check_for_dummy:
		if (HC_DUMMY == cmd)
			return 0;
		DRM_ERROR("Illegal DMA HC_DUMMY command\n");
		break;
	case check_for_dd:
		if (0xdddddddd == cmd)
			return 0;
		DRM_ERROR("Illegal DMA 0xdddddddd command\n");
		break;
	case check_z_buffer_addr0:
		cur_seq->unfinished = z_address;
		cur_seq->z_addr = (cur_seq->z_addr & 0xFF000000) |
		    (cmd & 0x00FFFFFF);
		return 0;
	case check_z_buffer_addr1:
		cur_seq->unfinished = z_address;
		cur_seq->z_addr = (cur_seq->z_addr & 0x00FFFFFF) |
		    ((cmd & 0xFF) << 24);
		return 0;
	case check_z_buffer_addr_mode:
		cur_seq->unfinished = z_address;
		if ((cmd & 0x0000C000) == 0)
			return 0;
		DRM_ERROR("Attempt to place Z buffer in system memory\n");
		return 2;
	case check_destination_addr0:
		cur_seq->unfinished = dest_address;
		cur_seq->d_addr = (cur_seq->d_addr & 0xFF000000) |
		    (cmd & 0x00FFFFFF);
		return 0;
	case check_destination_addr1:
		cur_seq->unfinished = dest_address;
		cur_seq->d_addr = (cur_seq->d_addr & 0x00FFFFFF) |
		    ((cmd & 0xFF) << 24);
		return 0;
	case check_destination_addr_mode:
		cur_seq->unfinished = dest_address;
		if ((cmd & 0x0000C000) == 0)
			return 0;
		DRM_ERROR
		    ("Attempt to place 3D drawing buffer in system memory\n");
		return 2;
	case check_texture_addr0:
		cur_seq->unfinished = tex_address;
		tmp = (cmd >> 24);
		tmp_addr = &cur_seq->t_addr[cur_seq->texture][tmp];
		*tmp_addr = (*tmp_addr & 0xFF000000) | (cmd & 0x00FFFFFF);
		return 0;
	case check_texture_addr1:
		cur_seq->unfinished = tex_address;
		tmp = ((cmd >> 24) - 0x20);
		tmp += tmp << 1;
		tmp_addr = &cur_seq->t_addr[cur_seq->texture][tmp];
		*tmp_addr = (*tmp_addr & 0x00FFFFFF) | ((cmd & 0xFF) << 24);
		tmp_addr++;
		*tmp_addr = (*tmp_addr & 0x00FFFFFF) | ((cmd & 0xFF00) << 16);
		tmp_addr++;
		*tmp_addr = (*tmp_addr & 0x00FFFFFF) | ((cmd & 0xFF0000) << 8);
		return 0;
	case check_texture_addr2:
		cur_seq->unfinished = tex_address;
		cur_seq->tex_level_lo[tmp = cur_seq->texture] = cmd & 0x3F;
		cur_seq->tex_level_hi[tmp] = (cmd & 0xFC0) >> 6;
		return 0;
	case check_texture_addr3:
		cur_seq->unfinished = tex_address;
		tmp = ((cmd >> 24) - HC_SubA_HTXnL0Pit);
		if (tmp == 0 && 
		    (cmd & HC_HTXnEnPit_MASK)) {
			cur_seq->pitch[cur_seq->texture][tmp] = 
				(cmd & HC_HTXnLnPit_MASK);
			cur_seq->tex_npot[cur_seq->texture] = 1;
		} else {
			cur_seq->pitch[cur_seq->texture][tmp] =
				(cmd & HC_HTXnLnPitE_MASK) >> HC_HTXnLnPitE_SHIFT;
			cur_seq->tex_npot[cur_seq->texture] = 0;			
			if (cmd & 0x000FFFFF) {
				DRM_ERROR
					("Unimplemented texture level 0 pitch mode.\n");
				return 2;
			}
		}
		return 0;
	case check_texture_addr4:
		cur_seq->unfinished = tex_address;
		tmp_addr = &cur_seq->t_addr[cur_seq->texture][9];
		*tmp_addr = (*tmp_addr & 0x00FFFFFF) | ((cmd & 0xFF) << 24);
		return 0;
	case check_texture_addr5:
	case check_texture_addr6:
		cur_seq->unfinished = tex_address;
		/*
		 * Texture width. We don't care since we have the pitch.
		 */
		return 0;
	case check_texture_addr7:
		cur_seq->unfinished = tex_address;
		tmp_addr = &(cur_seq->height[cur_seq->texture][0]);
		tmp_addr[5] = 1 << ((cmd & 0x00F00000) >> 20);
		tmp_addr[4] = 1 << ((cmd & 0x000F0000) >> 16);
		tmp_addr[3] = 1 << ((cmd & 0x0000F000) >> 12);
		tmp_addr[2] = 1 << ((cmd & 0x00000F00) >> 8);
		tmp_addr[1] = 1 << ((cmd & 0x000000F0) >> 4);
		tmp_addr[0] = 1 << (cmd & 0x0000000F);
		return 0;
	case check_texture_addr8:
		cur_seq->unfinished = tex_address;
		tmp_addr = &(cur_seq->height[cur_seq->texture][0]);
		tmp_addr[9] = 1 << ((cmd & 0x0000F000) >> 12);
		tmp_addr[8] = 1 << ((cmd & 0x00000F00) >> 8);
		tmp_addr[7] = 1 << ((cmd & 0x000000F0) >> 4);
		tmp_addr[6] = 1 << (cmd & 0x0000000F);
		return 0;
	case check_texture_addr_mode:
		cur_seq->unfinished = tex_address;
		if (2 == (tmp = cmd & 0x00000003)) {
			DRM_ERROR
			    ("Attempt to fetch texture from system memory.\n");
			return 2;
		}
		cur_seq->agp_texture = (tmp == 3);
		cur_seq->tex_palette_size[cur_seq->texture] =
		    (cmd >> 16) & 0x000000007;
		return 0;
	case check_for_vertex_count:
		cur_seq->vertex_count = cmd & 0x0000FFFF;
		return 0;
	case check_number_texunits:
		cur_seq->multitex = (cmd >> 3) & 1;
		return 0;
	default:
		DRM_ERROR("Illegal DMA data: 0x%x\n", cmd);
		return 2;
	}
	return 2;
}

static __inline__ int
via_check_prim_list(uint32_t const **buffer, const uint32_t * buf_end,
		    drm_via_state_t * cur_seq)
{
	drm_via_private_t *dev_priv =
	    (drm_via_private_t *) cur_seq->dev->dev_private;
	uint32_t a_fire, bcmd, dw_count;
	int ret = 0;
	int have_fire;
	const uint32_t *buf = *buffer;

	while (buf < buf_end) {
		have_fire = 0;
		if ((buf_end - buf) < 2) {
			DRM_ERROR
			    ("Unexpected termination of primitive list.\n");
			ret = 1;
			break;
		}
		if ((*buf & HC_ACMD_MASK) != HC_ACMD_HCmdB)
			break;
		bcmd = *buf++;
		if ((*buf & HC_ACMD_MASK) != HC_ACMD_HCmdA) {
			DRM_ERROR("Expected Vertex List A command, got 0x%x\n",
				  *buf);
			ret = 1;
			break;
		}
		a_fire =
		    *buf++ | HC_HPLEND_MASK | HC_HPMValidN_MASK |
		    HC_HE3Fire_MASK;

		/*
		 * How many dwords per vertex ?
		 */

		if (cur_seq->agp && ((bcmd & (0xF << 11)) == 0)) {
			DRM_ERROR("Illegal B command vertex data for AGP.\n");
			ret = 1;
			break;
		}

		dw_count = 0;
		if (bcmd & (1 << 7))
			dw_count += (cur_seq->multitex) ? 2 : 1;
		if (bcmd & (1 << 8))
			dw_count += (cur_seq->multitex) ? 2 : 1;
		if (bcmd & (1 << 9))
			dw_count++;
		if (bcmd & (1 << 10))
			dw_count++;
		if (bcmd & (1 << 11))
			dw_count++;
		if (bcmd & (1 << 12))
			dw_count++;
		if (bcmd & (1 << 13))
			dw_count++;
		if (bcmd & (1 << 14))
			dw_count++;

		while (buf < buf_end) {
			if (*buf == a_fire) {
				if (dev_priv->num_fire_offsets >=
				    VIA_FIRE_BUF_SIZE) {
					DRM_ERROR("Fire offset buffer full.\n");
					ret = 1;
					break;
				}
				dev_priv->fire_offsets[dev_priv->
						       num_fire_offsets++] =
				    buf;
				have_fire = 1;
				buf++;
				if (buf < buf_end && *buf == a_fire)
					buf++;
				break;
			}
			if ((*buf == HALCYON_HEADER2) ||
			    ((*buf & HALCYON_FIREMASK) == HALCYON_FIRECMD)) {
				DRM_ERROR("Missing Vertex Fire command, "
					  "Stray Vertex Fire command  or verifier "
					  "lost sync.\n");
				ret = 1;
				break;
			}
			if ((ret = eat_words(&buf, buf_end, dw_count)))
				break;
		}
		if (buf >= buf_end && !have_fire) {
			DRM_ERROR("Missing Vertex Fire command or verifier "
				  "lost sync.\n");
			ret = 1;
			break;
		}
		if (cur_seq->agp && ((buf - cur_seq->buf_start) & 0x01)) {
			DRM_ERROR("AGP Primitive list end misaligned.\n");
			ret = 1;
			break;
		}
	}
	*buffer = buf;
	return ret;
}

static __inline__ verifier_state_t
via_check_header2(uint32_t const **buffer, const uint32_t * buf_end,
		  drm_via_state_t * hc_state)
{
	uint32_t cmd;
	int hz_mode;
	hazard_t hz;
	const uint32_t *buf = *buffer;
	const hazard_t *hz_table;

	if ((buf_end - buf) < 2) {
		DRM_ERROR
		    ("Illegal termination of DMA HALCYON_HEADER2 sequence.\n");
		return state_error;
	}
	buf++;
	cmd = (*buf++ & 0xFFFF0000) >> 16;

	switch (cmd) {
	case HC_ParaType_CmdVdata:
		if (via_check_prim_list(&buf, buf_end, hc_state))
			return state_error;
		*buffer = buf;
		return state_command;
	case HC_ParaType_NotTex:
		hz_table = table1;
		break;
	case HC_ParaType_Tex:
		hc_state->texture = 0;
		hz_table = table2;
		break;
	case (HC_ParaType_Tex | (HC_SubType_Tex1 << 8)):
		hc_state->texture = 1;
		hz_table = table2;
		break;
	case (HC_ParaType_Tex | (HC_SubType_TexGeneral << 8)):
		hz_table = table3;
		break;
	case HC_ParaType_Auto:
		if (eat_words(&buf, buf_end, 2))
			return state_error;
		*buffer = buf;
		return state_command;
	case (HC_ParaType_Palette | (HC_SubType_Stipple << 8)):
		if (eat_words(&buf, buf_end, 32))
			return state_error;
		*buffer = buf;
		return state_command;
	case (HC_ParaType_Palette | (HC_SubType_TexPalette0 << 8)):
	case (HC_ParaType_Palette | (HC_SubType_TexPalette1 << 8)):
		DRM_ERROR("Texture palettes are rejected because of "
			  "lack of info how to determine their size.\n");
		return state_error;
	case (HC_ParaType_Palette | (HC_SubType_FogTable << 8)):
		DRM_ERROR("Fog factor palettes are rejected because of "
			  "lack of info how to determine their size.\n");
		return state_error;
	default:

		/*
		 * There are some unimplemented HC_ParaTypes here, that
		 * need to be implemented if the Mesa driver is extended.
		 */

		DRM_ERROR("Invalid or unimplemented HALCYON_HEADER2 "
			  "DMA subcommand: 0x%x. Previous dword: 0x%x\n",
			  cmd, *(buf - 2));
		*buffer = buf;
		return state_error;
	}

	while (buf < buf_end) {
		cmd = *buf++;
		if ((hz = hz_table[cmd >> 24])) {
			if ((hz_mode = investigate_hazard(cmd, hz, hc_state))) {
				if (hz_mode == 1) {
					buf--;
					break;
				}
				return state_error;
			}
		} else if (hc_state->unfinished &&
			   finish_current_sequence(hc_state)) {
			return state_error;
		}
	}
	if (hc_state->unfinished && finish_current_sequence(hc_state)) {
		return state_error;
	}
	*buffer = buf;
	return state_command;
}

static __inline__ verifier_state_t
via_parse_header2(drm_via_private_t * dev_priv, uint32_t const **buffer,
		  const uint32_t * buf_end, int *fire_count)
{
	uint32_t cmd;
	const uint32_t *buf = *buffer;
	const uint32_t *next_fire;
	int burst = 0;

	next_fire = dev_priv->fire_offsets[*fire_count];
	buf++;
	cmd = (*buf & 0xFFFF0000) >> 16;
	VIA_WRITE(HC_REG_TRANS_SET + HC_REG_BASE, *buf++);
	switch (cmd) {
	case HC_ParaType_CmdVdata:
		while ((buf < buf_end) &&
		       (*fire_count < dev_priv->num_fire_offsets) &&
		       (*buf & HC_ACMD_MASK) == HC_ACMD_HCmdB) {
			while (buf <= next_fire) {
				VIA_WRITE(HC_REG_TRANS_SPACE + HC_REG_BASE +
					  (burst & 63), *buf++);
				burst += 4;
			}
			if ((buf < buf_end)
			    && ((*buf & HALCYON_FIREMASK) == HALCYON_FIRECMD))
				buf++;

			if (++(*fire_count) < dev_priv->num_fire_offsets)
				next_fire = dev_priv->fire_offsets[*fire_count];
		}
		break;
	default:
		while (buf < buf_end) {

			if (*buf == HC_HEADER2 ||
			    (*buf & HALCYON_HEADER1MASK) == HALCYON_HEADER1 ||
			    (*buf & VIA_VIDEOMASK) == VIA_VIDEO_HEADER5 ||
			    (*buf & VIA_VIDEOMASK) == VIA_VIDEO_HEADER6)
				break;

			VIA_WRITE(HC_REG_TRANS_SPACE + HC_REG_BASE +
				  (burst & 63), *buf++);
			burst += 4;
		}
	}
	*buffer = buf;
	return state_command;
}

static __inline__ int verify_mmio_address(uint32_t address)
{
	if ((address > 0x3FF) && (address < 0xC00)) {
		DRM_ERROR("Invalid VIDEO DMA command. "
			  "Attempt to access 3D- or command burst area.\n");
		return 1;
	} else if ((address > 0xCFF) && (address < 0x1300)) {
		DRM_ERROR("Invalid VIDEO DMA command. "
			  "Attempt to access PCI DMA area.\n");
		return 1;
	} else if (address > 0x13FF) {
		DRM_ERROR("Invalid VIDEO DMA command. "
			  "Attempt to access VGA registers.\n");
		return 1;
	}
	return 0;
}

static __inline__ int
verify_video_tail(uint32_t const **buffer, const uint32_t * buf_end,
		  uint32_t dwords)
{
	const uint32_t *buf = *buffer;

	if (buf_end - buf < dwords) {
		DRM_ERROR("Illegal termination of video command.\n");
		return 1;
	}
	while (dwords--) {
		if (*buf++) {
			DRM_ERROR("Illegal video command tail.\n");
			return 1;
		}
	}
	*buffer = buf;
	return 0;
}

static __inline__ verifier_state_t
via_check_header1(uint32_t const **buffer, const uint32_t * buf_end)
{
	uint32_t cmd;
	const uint32_t *buf = *buffer;
	verifier_state_t ret = state_command;

	while (buf < buf_end) {
		cmd = *buf;
		if ((cmd > ((0x3FF >> 2) | HALCYON_HEADER1)) &&
		    (cmd < ((0xC00 >> 2) | HALCYON_HEADER1))) {
			if ((cmd & HALCYON_HEADER1MASK) != HALCYON_HEADER1)
				break;
			DRM_ERROR("Invalid HALCYON_HEADER1 command. "
				  "Attempt to access 3D- or command burst area.\n");
			ret = state_error;
			break;
		} else if (cmd > ((0xCFF >> 2) | HALCYON_HEADER1)) {
			if ((cmd & HALCYON_HEADER1MASK) != HALCYON_HEADER1)
				break;
			DRM_ERROR("Invalid HALCYON_HEADER1 command. "
				  "Attempt to access VGA registers.\n");
			ret = state_error;
			break;
		} else {
			buf += 2;
		}
	}
	*buffer = buf;
	return ret;
}

static __inline__ verifier_state_t
via_parse_header1(drm_via_private_t * dev_priv, uint32_t const **buffer,
		  const uint32_t * buf_end)
{
	register uint32_t cmd;
	const uint32_t *buf = *buffer;

	while (buf < buf_end) {
		cmd = *buf;
		if ((cmd & HALCYON_HEADER1MASK) != HALCYON_HEADER1)
			break;
		VIA_WRITE((cmd & ~HALCYON_HEADER1MASK) << 2, *++buf);
		buf++;
	}
	*buffer = buf;
	return state_command;
}

static __inline__ verifier_state_t
via_check_vheader5(uint32_t const **buffer, const uint32_t * buf_end)
{
	uint32_t data;
	const uint32_t *buf = *buffer;

	if (buf_end - buf < 4) {
		DRM_ERROR("Illegal termination of video header5 command\n");
		return state_error;
	}

	data = *buf++ & ~VIA_VIDEOMASK;
	if (verify_mmio_address(data))
		return state_error;

	data = *buf++;
	if (*buf++ != 0x00F50000) {
		DRM_ERROR("Illegal header5 header data\n");
		return state_error;
	}
	if (*buf++ != 0x00000000) {
		DRM_ERROR("Illegal header5 header data\n");
		return state_error;
	}
	if (eat_words(&buf, buf_end, data))
		return state_error;
	if ((data & 3) && verify_video_tail(&buf, buf_end, 4 - (data & 3)))
		return state_error;
	*buffer = buf;
	return state_command;

}

static __inline__ verifier_state_t
via_parse_vheader5(drm_via_private_t * dev_priv, uint32_t const **buffer,
		   const uint32_t * buf_end)
{
	uint32_t addr, count, i;
	const uint32_t *buf = *buffer;

	addr = *buf++ & ~VIA_VIDEOMASK;
	i = count = *buf;
	buf += 3;
	while (i--) {
		VIA_WRITE(addr, *buf++);
	}
	if (count & 3)
		buf += 4 - (count & 3);
	*buffer = buf;
	return state_command;
}

static __inline__ verifier_state_t
via_check_vheader6(uint32_t const **buffer, const uint32_t * buf_end)
{
	uint32_t data;
	const uint32_t *buf = *buffer;
	uint32_t i;

	if (buf_end - buf < 4) {
		DRM_ERROR("Illegal termination of video header6 command\n");
		return state_error;
	}
	buf++;
	data = *buf++;
	if (*buf++ != 0x00F60000) {
		DRM_ERROR("Illegal header6 header data\n");
		return state_error;
	}
	if (*buf++ != 0x00000000) {
		DRM_ERROR("Illegal header6 header data\n");
		return state_error;
	}
	if ((buf_end - buf) < (data << 1)) {
		DRM_ERROR("Illegal termination of video header6 command\n");
		return state_error;
	}
	for (i = 0; i < data; ++i) {
		if (verify_mmio_address(*buf++))
			return state_error;
		buf++;
	}
	data <<= 1;
	if ((data & 3) && verify_video_tail(&buf, buf_end, 4 - (data & 3)))
		return state_error;
	*buffer = buf;
	return state_command;
}

static __inline__ verifier_state_t
via_parse_vheader6(drm_via_private_t * dev_priv, uint32_t const **buffer,
		   const uint32_t * buf_end)
{

	uint32_t addr, count, i;
	const uint32_t *buf = *buffer;

	i = count = *++buf;
	buf += 3;
	while (i--) {
		addr = *buf++;
		VIA_WRITE(addr, *buf++);
	}
	count <<= 1;
	if (count & 3)
		buf += 4 - (count & 3);
	*buffer = buf;
	return state_command;
}

int
via_verify_command_stream(const uint32_t * buf, unsigned int size,
			  drm_device_t * dev, int agp)
{

	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
	drm_via_state_t *hc_state = &dev_priv->hc_state;
	drm_via_state_t saved_state = *hc_state;
	uint32_t cmd;
	const uint32_t *buf_end = buf + (size >> 2);
	verifier_state_t state = state_command;
	int cme_video;
	int supported_3d;

	cme_video = (dev_priv->chipset == VIA_PRO_GROUP_A ||
		     dev_priv->chipset == VIA_DX9_0);

	supported_3d = dev_priv->chipset != VIA_DX9_0;

	hc_state->dev = dev;
	hc_state->unfinished = no_sequence;
	hc_state->map_cache = NULL;
	hc_state->agp = agp;
	hc_state->buf_start = buf;
	dev_priv->num_fire_offsets = 0;

	while (buf < buf_end) {

		switch (state) {
		case state_header2:
			state = via_check_header2(&buf, buf_end, hc_state);
			break;
		case state_header1:
			state = via_check_header1(&buf, buf_end);
			break;
		case state_vheader5:
			state = via_check_vheader5(&buf, buf_end);
			break;
		case state_vheader6:
			state = via_check_vheader6(&buf, buf_end);
			break;
		case state_command:
			if ((HALCYON_HEADER2 == (cmd = *buf)) && 
			    supported_3d)
				state = state_header2;
			else if ((cmd & HALCYON_HEADER1MASK) == HALCYON_HEADER1)
				state = state_header1;
			else if (cme_video
				 && (cmd & VIA_VIDEOMASK) == VIA_VIDEO_HEADER5)
				state = state_vheader5;
			else if (cme_video
				 && (cmd & VIA_VIDEOMASK) == VIA_VIDEO_HEADER6)
				state = state_vheader6;
			else if ((cmd == HALCYON_HEADER2) && !supported_3d) {
				DRM_ERROR("Accelerated 3D is not supported on this chipset yet.\n");
				state = state_error;
			} else {
				DRM_ERROR
				    ("Invalid / Unimplemented DMA HEADER command. 0x%x\n",
				     cmd);
				state = state_error;
			}
			break;
		case state_error:
		default:
			*hc_state = saved_state;
			return DRM_ERR(EINVAL);
		}
	}
	if (state == state_error) {
		*hc_state = saved_state;
		return DRM_ERR(EINVAL);
	}
	return 0;
}

int
via_parse_command_stream(drm_device_t * dev, const uint32_t * buf,
			 unsigned int size)
{

	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
	uint32_t cmd;
	const uint32_t *buf_end = buf + (size >> 2);
	verifier_state_t state = state_command;
	int fire_count = 0;

	while (buf < buf_end) {

		switch (state) {
		case state_header2:
			state =
			    via_parse_header2(dev_priv, &buf, buf_end,
					      &fire_count);
			break;
		case state_header1:
			state = via_parse_header1(dev_priv, &buf, buf_end);
			break;
		case state_vheader5:
			state = via_parse_vheader5(dev_priv, &buf, buf_end);
			break;
		case state_vheader6:
			state = via_parse_vheader6(dev_priv, &buf, buf_end);
			break;
		case state_command:
			if (HALCYON_HEADER2 == (cmd = *buf))
				state = state_header2;
			else if ((cmd & HALCYON_HEADER1MASK) == HALCYON_HEADER1)
				state = state_header1;