smatch_kernel_host_data.c

   1 /*
   2  * Copyright (C) 2011 Dan Carpenter.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License
   6  * as published by the Free Software Foundation; either version 2
   7  * of the License, or (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, see http://www.gnu.org/copyleft/gpl.txt
  16  */
  17
  18 /* The below code is an adjustment of the smatch_kernel_user_data.c to
  19  * track (untrusted and potentially malicious) data received by a guest
  20  * kernel via various low-level port IO, MMIO, MSR, CPUID and PCI config
  21  * space access functions (see func_table for a full list) from an
  22  * untrusted host/VMM under confidential computing threat model
  23  * (where a guest VM does not trust the host/VMM).
  24  * We call this data as 'host' data here. Being able to track host data
  25  * allows creating new smatch patterns that can perform various checks
  26  * on such data, i.e. verify that that there are no spectre v1 gadgets
  27  * on this attack surface or even simply report locations where host data
  28  * is being processed in the kernel source code (useful for a targeted code
  29  * audit).
  30  * Similar as smatch_kernel_user_data.c it works with
  31  * smatch_points_to_host_input.c code.
  32  *
  33  * Copyright (C) 2022 Elena Reshetova Intel Corporation.
  34  */
  35
  36 #include "smatch.h"
  37 #include "smatch_slist.h"
  38 #include "smatch_extra.h"
  39
  40 struct host_fn_info {
  41         const char *name;
  42         int type;
  43         int param;
  44         const char *key;
  45         const sval_t *implies_start, *implies_end;
  46         func_hook *call_back;
  47 };
  48
  49 static struct host_fn_info func_table[] = {
  50                 /* Functions that return host data as return value */
  51             { "inb", HOST_DATA, -1, "$" },
  52                 { "inw", HOST_DATA, -1, "$" },
  53             { "inl", HOST_DATA, -1, "$" },
  54                 { "__inb", HOST_DATA, -1, "$" },
  55                 { "__inw", HOST_DATA, -1, "$" },
  56             { "__inl", HOST_DATA, -1, "$" },
  57             { "ioread8", HOST_DATA, -1, "$" },
  58             { "ioread16", HOST_DATA, -1, "$" },
  59             { "ioread32", HOST_DATA, -1, "$" },
  60             { "ioread16be", HOST_DATA, -1, "$" },
  61             { "ioread32be", HOST_DATA, -1, "$" },
  62             { "ioread64_lo_hi", HOST_DATA, -1, "$" },
  63             { "ioread64_hi_lo", HOST_DATA, -1, "$" },
  64             { "ioread64be_lo_hi", HOST_DATA, -1, "$" },
  65             { "ioread64be_hi_lo", HOST_DATA, -1, "$" },
  66             { "iomap_readq", HOST_DATA, -1, "$" },
  67             { "iomap_readb", HOST_DATA, -1, "$" },
  68             { "iomap_readw", HOST_DATA, -1, "$" },
  69             { "iomap_readl", HOST_DATA, -1, "$" },
  70             { "readb", HOST_DATA, -1, "$" },
  71             { "readw", HOST_DATA, -1, "$" },
  72             { "readl", HOST_DATA, -1, "$" },
  73             { "readq", HOST_DATA, -1, "$" },
  74             { "__raw_readb", HOST_DATA, -1, "$" },
  75             { "__raw_readw", HOST_DATA, -1, "$" },
  76             { "__raw_readl", HOST_DATA, -1, "$" },
  77                 { "__raw_readq", HOST_DATA, -1, "$" },
  78             { "__readq", HOST_DATA, -1, "$" },
  79             { "__readl", HOST_DATA, -1, "$" },
  80             { "__readb", HOST_DATA, -1, "$" },
  81             { "__readw", HOST_DATA, -1, "$" },
  82             { "native_read_msr", HOST_DATA, -1, "$" },
  83             { "native_read_msr_safe", HOST_DATA, -1, "$" },
  84             { "__rdmsr", HOST_DATA, -1, "$" },
  85             { "paravirt_read_msr", HOST_DATA, -1, "$" },
  86             { "paravirt_read_msr_safe", HOST_DATA, -1, "$" },
  87             /* the below 3 apic funcs are needed
  88              * since they unwrap to apic->read(reg),
  89              * apic->icr_read() and *(APIC_BASE + reg)*/
  90             { "apic_read", HOST_DATA, -1, "$" },
  91             { "apic_icr_read", HOST_DATA, -1, "$" },
  92             { "native_apic_mem_read", HOST_DATA, -1, "$" },
  93             /* the below one is x86_apic_ops.io_apic_read */
  94             { "io_apic_read", HOST_DATA, -1, "$" },
  95             /* virtio_cread8/16/32/64 funcs are needed
  96              * since they call vdev->config->get */
  97             { "virtio_cread8", HOST_DATA, -1, "$" },
  98             { "virtio_cread16", HOST_DATA, -1, "$" },
  99             { "virtio_cread32", HOST_DATA, -1, "$" },
 100             { "virtio_cread64", HOST_DATA, -1, "$" },
 101             { "__virtio16_to_cpu", HOST_DATA, -1, "$" },
 102             { "__virtio32_to_cpu", HOST_DATA, -1, "$" },
 103             { "__virtio64_to_cpu", HOST_DATA, -1, "$" },
 104             { "serial_dl_read", HOST_DATA, -1, "$" },
 105             { "serial_in", HOST_DATA, -1, "$" },
 106             { "serial_port_in", HOST_DATA, -1, "$" },
 107             { "cpuid_eax", HOST_DATA, -1, "$" },
 108             { "cpuid_ebx", HOST_DATA, -1, "$" },
 109             { "cpuid_ecx", HOST_DATA, -1, "$" },
 110             { "cpuid_edx", HOST_DATA, -1, "$" },
 111             /* Functions that return host data as argument 1 */
 112         { "memcpy_fromio", HOST_DATA, 0, "*$" },
 113             /* Functions that return host data as argument 2 */
 114         { "acpi_os_read_iomem", HOST_DATA, 1, "*$" },
 115         { "mmio_insb", HOST_DATA, 1, "*$" },
 116         { "mmio_insw", HOST_DATA, 1, "*$" },
 117         { "mmio_insl", HOST_DATA, 1, "*$" },
 118         { "acpi_read_bit_register", HOST_DATA, 1, "*$" },
 119           /* Functions that return host data as argument 3 */
 120         { "rdmsrl_on_cpu", HOST_DATA, 2, "*$" },
 121         { "rdmsrl_safe_on_cpu", HOST_DATA, 2, "*$" },
 122         { "__virtio_cread_many", HOST_DATA, 2, "*$" },
 123         { "pci_user_read_config_word", HOST_DATA, 2, "*$" },
 124         { "pci_user_read_config_dword", HOST_DATA, 2, "*$" },
 125         { "pci_user_read_config_byte", HOST_DATA, 2, "*$" },
 126           /* Functions that return host data as argument 4 */
 127         { "pci_bus_read_config_byte", HOST_DATA, 3, "*$" },
 128         { "pci_bus_read_config_word", HOST_DATA, 3, "*$" },
 129         { "pci_bus_read_config_dword", HOST_DATA, 3, "*$" },
 130          /* Functions that return host data as arguments 3 and 4 */
 131         { "rdmsr_on_cpu", HOST_DATA, 2, "*$" },
 132         { "rdmsr_on_cpu", HOST_DATA, 3, "*$" },
 133         { "rdmsr_safe_on_cpu", HOST_DATA, 2, "*$" },
 134         { "rdmsr_safe_on_cpu", HOST_DATA, 3, "*$" },
 135         /* Functions that return host data as argument 6 */
 136         { "raw_pci_read", HOST_DATA, 5, "*$" },
 137         /* Functions that return host data as arguments 2-5 */
 138         { "cpuid", HOST_DATA, 1, "*$" },
 139         { "cpuid", HOST_DATA, 2, "*$" },
 140         { "cpuid", HOST_DATA, 3, "*$" },
 141         { "cpuid", HOST_DATA, 4, "*$" },
 142         /* Functions that return host data as arguments 3-6 */
 143         { "cpuid_count", HOST_DATA, 2, "*$" },
 144         { "cpuid_count", HOST_DATA, 3, "*$" },
 145         { "cpuid_count", HOST_DATA, 4, "*$" },
 146         { "cpuid_count", HOST_DATA, 5, "*$" },
 147 };
 148
 149 static int my_id;
 150 static unsigned long func_gets_host_data;
 151 static struct stree *start_states;
 152
 153 static struct smatch_state *empty_state(struct sm_state *sm)
 154 {
 155         return alloc_estate_empty();
 156 }
 157
 158 static struct smatch_state *new_state(struct symbol *type)
 159 {
 160         struct smatch_state *state;
 161
 162         if (!type || type_is_ptr(type))
 163                 return NULL;
 164
 165         state = alloc_estate_whole(type);
 166         estate_set_new(state);
 167         return state;
 168 }
 169
 170 static void pre_merge_hook(struct sm_state *cur, struct sm_state *other)
 171 {
 172         struct smatch_state *kernel = cur->state;
 173         struct smatch_state *extra;
 174         struct smatch_state *state;
 175         struct range_list *rl;
 176
 177         extra = __get_state(SMATCH_EXTRA, cur->name, cur->sym);
 178         if (!extra)
 179                 return;
 180         rl = rl_intersection(estate_rl(kernel), estate_rl(extra));
 181         state = alloc_estate_rl(clone_rl(rl));
 182         if (estate_capped(kernel) || is_capped_var_sym(cur->name, cur->sym))
 183                 estate_set_capped(state);
 184         if (estates_equiv(state, cur->state))
 185                 return;
 186         if (estate_new(cur->state))
 187                 estate_set_new(state);
 188         set_state(my_id, cur->name, cur->sym, state);
 189 }
 190
 191 static void extra_nomod_hook(const char *name, struct symbol *sym, struct expression *expr, struct smatch_state *state)
 192 {
 193         struct smatch_state *host, *new;
 194         struct range_list *rl;
 195
 196         host = __get_state(my_id, name, sym);
 197         if (!host)
 198                 return;
 199
 200         rl = rl_intersection(estate_rl(host), estate_rl(state));
 201         if (rl_equiv(rl, estate_rl(host)))
 202                 return;
 203         new = alloc_estate_rl(rl);
 204         if (estate_capped(host))
 205                 estate_set_capped(new);
 206         set_state(my_id, name, sym, new);
 207 }
 208
 209 static void store_type_info(struct expression *expr, struct smatch_state *state)
 210 {
 211         struct symbol *type;
 212         char *type_str, *member;
 213
 214         if (__in_fake_assign)
 215                 return;
 216
 217         if (!estate_rl(state))
 218                 return;
 219
 220         expr = strip_expr(expr);
 221         if (!expr || expr->type != EXPR_DEREF || !expr->member)
 222                 return;
 223
 224         type = get_type(expr->deref);
 225         if (!type || !type->ident)
 226                 return;
 227
 228         type_str = type_to_str(type);
 229         if (!type_str)
 230                 return;
 231         member = get_member_name(expr);
 232         if (!member)
 233                 return;
 234
 235         sql_insert_function_type_info(HOST_DATA, type_str, member, state->name);
 236 }
 237
 238 static void set_host_data(struct expression *expr, struct smatch_state *state)
 239 {
 240         store_type_info(expr, state);
 241         set_state_expr(my_id, expr, state);
 242 }
 243
 244 static bool host_rl_known(struct expression *expr)
 245 {
 246         struct range_list *rl;
 247         sval_t close_to_max;
 248
 249         if (!get_host_rl(expr, &rl))
 250                 return true;
 251
 252         close_to_max = sval_type_max(rl_type(rl));
 253         close_to_max.value -= 100;
 254
 255         if (sval_cmp(rl_max(rl), close_to_max) >= 0)
 256                 return false;
 257         return true;
 258 }
 259
 260 static bool is_array_index_mask_nospec(struct expression *expr)
 261 {
 262         struct expression *orig;
 263
 264         orig = get_assigned_expr(expr);
 265         if (!orig || orig->type != EXPR_CALL)
 266                 return false;
 267         return sym_name_is("array_index_mask_nospec", orig->fn);
 268 }
 269
 270 static bool binop_capped(struct expression *expr)
 271 {
 272         struct range_list *left_rl;
 273         int comparison;
 274         sval_t sval;
 275
 276         if (expr->op == '-' && get_host_rl(expr->left, &left_rl)) {
 277                 if (host_rl_capped(expr->left))
 278                         return true;
 279                 comparison = get_comparison(expr->left, expr->right);
 280                 if (comparison && show_special(comparison)[0] == '>')
 281                         return true;
 282                 return false;
 283         }
 284
 285         if (expr->op == '&' || expr->op == '%') {
 286                 bool left_host, left_capped, right_host, right_capped;
 287
 288                 if (!get_value(expr->right, &sval) && is_capped(expr->right))
 289                         return true;
 290                 if (is_array_index_mask_nospec(expr->right))
 291                         return true;
 292                 if (is_capped(expr->left))
 293                         return true;
 294                 left_host = is_host_rl(expr->left);
 295                 right_host = is_host_rl(expr->right);
 296                 if (!left_host && !right_host)
 297                         return true;
 298
 299                 left_capped = host_rl_capped(expr->left);
 300                 right_capped = host_rl_capped(expr->right);
 301
 302                 if (left_host && left_capped) {
 303                         if (!right_host)
 304                                 return true;
 305                         if (right_host && right_capped)
 306                                 return true;
 307                         return false;
 308                 }
 309                 if (right_host && right_capped) {
 310                         if (!left_host)
 311                                 return true;
 312                         return false;
 313                 }
 314                 return false;
 315         }
 316
 317         /*
 318          * Generally "capped" means that we capped it to an unknown value.
 319          * This is useful because if Smatch doesn't know what the value is then
 320          * we have to trust that it is correct.  But if we known cap value is
 321          * 100 then we can check if 100 is correct and complain if it's wrong.
 322          *
 323          * So then the problem is with BINOP when we take a capped variable
 324          * plus a host variable which is clamped to a known range (uncapped)
 325          * the result should be capped.
 326          */
 327         if ((host_rl_capped(expr->left) || host_rl_known(expr->left)) &&
 328             (host_rl_capped(expr->right) || host_rl_known(expr->right)))
 329                 return true;
 330         return false;
 331 }
 332
 333 bool host_rl_capped(struct expression *expr)
 334 {
 335         struct smatch_state *state;
 336         struct range_list *rl;
 337         sval_t sval;
 338
 339         expr = strip_expr(expr);
 340         if (!expr)
 341                 return false;
 342         if (get_value(expr, &sval))
 343                 return true;
 344         if (expr->type == EXPR_BINOP)
 345                 return binop_capped(expr);
 346         if ((expr->type == EXPR_PREOP || expr->type == EXPR_POSTOP) &&
 347             (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT))
 348                 return host_rl_capped(expr->unop);
 349         state = get_state_expr(my_id, expr);
 350         if (state)
 351                 return estate_capped(state);
 352
 353         if (!get_host_rl(expr, &rl)) {
 354                 /*
 355                  * The non host data parts of a binop are capped and
 356                  * also empty host rl states are capped.
 357                  */
 358                 return true;
 359         }
 360
 361         if (rl_to_sval(rl, &sval))
 362                 return true;
 363
 364         return false;  /* uncapped host data */
 365 }
 366
 367 static void tag_inner_struct_members(struct expression *expr, struct symbol *member)
 368 {
 369         struct expression *edge_member;
 370         struct symbol *base = get_real_base_type(member);
 371         struct symbol *tmp;
 372
 373         if (member->ident)
 374                 expr = member_expression(expr, '.', member->ident);
 375
 376         FOR_EACH_PTR(base->symbol_list, tmp) {
 377                 struct symbol *type;
 378
 379                 type = get_real_base_type(tmp);
 380                 if (!type)
 381                         continue;
 382
 383                 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
 384                         tag_inner_struct_members(expr, tmp);
 385                         continue;
 386                 }
 387
 388                 if (!tmp->ident)
 389                         continue;
 390
 391                 edge_member = member_expression(expr, '.', tmp->ident);
 392                 set_host_data(edge_member, new_state(type));
 393         } END_FOR_EACH_PTR(tmp);
 394 }
 395
 396 static void tag_struct_members(struct symbol *type, struct expression *expr)
 397 {
 398         struct symbol *tmp;
 399         struct expression *member;
 400         int op = '*';
 401
 402         if (expr->type == EXPR_PREOP && expr->op == '&') {
 403                 expr = strip_expr(expr->unop);
 404                 op = '.';
 405         }
 406
 407         FOR_EACH_PTR(type->symbol_list, tmp) {
 408                 type = get_real_base_type(tmp);
 409                 if (!type)
 410                         continue;
 411
 412                 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
 413                         tag_inner_struct_members(expr, tmp);
 414                         continue;
 415                 }
 416
 417                 if (!tmp->ident)
 418                         continue;
 419
 420                 member = member_expression(expr, op, tmp->ident);
 421                 if (type->type == SYM_ARRAY) {
 422                         set_points_to_host_data(member);
 423                 } else {
 424                         set_host_data(member, new_state(get_type(member)));
 425                 }
 426         } END_FOR_EACH_PTR(tmp);
 427 }
 428
 429 static void tag_base_type(struct expression *expr)
 430 {
 431         if (expr->type == EXPR_PREOP && expr->op == '&')
 432                 expr = strip_expr(expr->unop);
 433         else
 434                 expr = deref_expression(expr);
 435         set_host_data(expr, new_state(get_type(expr)));
 436 }
 437
 438 static void tag_as_host_data(struct expression *expr)
 439 {
 440         struct symbol *type;
 441
 442         expr = strip_expr(expr);
 443         type = get_type(expr);
 444
 445         if (!type)
 446                 return;
 447
 448         if (type->type == SYM_PTR){
 449                 type = get_real_base_type(type);
 450                 if (!type)
 451                         return;
 452                 if (type == &void_ctype) {
 453                         set_host_data(deref_expression(expr), new_state(&ulong_ctype));
 454                         return;
 455                 }
 456                 if (type->type == SYM_BASETYPE) {
 457                         if (expr->type != EXPR_PREOP && expr->op != '&')
 458                                 set_points_to_host_data(expr);
 459                         tag_base_type(expr);
 460                         return;
 461                 }
 462                 if (type->type == SYM_STRUCT || type->type == SYM_UNION) {
 463                         if (expr->type != EXPR_PREOP || expr->op != '&')
 464                                 expr = deref_expression(expr);
 465                         else
 466                                 set_host_data(deref_expression(expr), new_state(&ulong_ctype));
 467                         tag_struct_members(type, expr);
 468                 }
 469                 return;
 470         }
 471
 472         if (expr->type == EXPR_DEREF) {
 473                 type = get_type(expr->deref);
 474                 expr = strip_expr(expr->deref);
 475
 476                 if (!type){
 477                         sm_msg("%s: no type\n", __func__);
 478                         return;
 479                 }
 480         }
 481
 482         if (type->type == SYM_BASETYPE) {
 483                 if (expr->type == EXPR_PREOP && expr->op == '*')
 484                         set_points_to_host_data(expr->unop);
 485                 set_host_data(expr, new_state(get_type(expr)));
 486                 return;
 487         }
 488         if (type->type == SYM_STRUCT || type->type == SYM_UNION) {
 489                 set_host_data(expr, new_state(&ulong_ctype));
 490                 tag_struct_members(type, expr);
 491         }
 492 }
 493
 494 static int get_rl_from_function(struct expression *expr, struct range_list **rl)
 495 {
 496         if (!expr)
 497                 return 0;
 498
 499         if (expr->type != EXPR_CALL || expr->fn->type != EXPR_SYMBOL ||
 500             !expr->fn->symbol_name || !expr->fn->symbol_name->name)
 501                 return 0;
 502
 503         for (int i = 0; i < ARRAY_SIZE(func_table); i++) {
 504                 if ((strcmp(expr->fn->symbol_name->name, func_table[i].name) == 0) &&
 505                    (func_table[i].param == -1)) {
 506                         *rl = alloc_whole_rl(get_type(expr));
 507                         return 1;
 508                 }
 509         }
 510         return 0;
 511 }
 512
 513 static bool state_is_new(struct expression *expr)
 514 {
 515         struct smatch_state *state;
 516
 517         state = get_state_expr(my_id, expr);
 518         if (estate_new(state))
 519                 return true;
 520
 521         if (expr->type == EXPR_BINOP) {
 522                 if (state_is_new(expr->left))
 523                         return true;
 524                 if (state_is_new(expr->right))
 525                         return true;
 526         }
 527         return false;
 528 }
 529
 530 static void handle_derefed_pointers(struct expression *expr)
 531 {
 532         expr = strip_expr(expr);
 533         if (expr->type != EXPR_PREOP ||
 534             expr->op != '*')
 535                 return;
 536         expr = strip_expr(expr->unop);
 537         set_points_to_host_data(expr);
 538 }
 539
 540 static bool handle_op_assign(struct expression *expr)
 541 {
 542         struct expression *binop_expr;
 543         struct smatch_state *state;
 544         struct range_list *rl;
 545
 546         switch (expr->op) {
 547         case SPECIAL_ADD_ASSIGN:
 548         case SPECIAL_SUB_ASSIGN:
 549         case SPECIAL_AND_ASSIGN:
 550         case SPECIAL_MOD_ASSIGN:
 551         case SPECIAL_SHL_ASSIGN:
 552         case SPECIAL_SHR_ASSIGN:
 553         case SPECIAL_OR_ASSIGN:
 554         case SPECIAL_XOR_ASSIGN:
 555         case SPECIAL_MUL_ASSIGN:
 556         case SPECIAL_DIV_ASSIGN:
 557                 binop_expr = binop_expression(expr->left,
 558                                               op_remove_assign(expr->op),
 559                                               expr->right);
 560                 if (!get_host_rl(binop_expr, &rl))
 561                         return true;
 562                 rl = cast_rl(get_type(expr->left), rl);
 563                 state = alloc_estate_rl(rl);
 564                 if (expr->op == SPECIAL_AND_ASSIGN ||
 565                     expr->op == SPECIAL_MOD_ASSIGN ||
 566                     host_rl_capped(binop_expr))
 567                         estate_set_capped(state);
 568                 if (state_is_new(binop_expr))
 569                         estate_set_new(state);
 570                 set_host_data(expr->left, state);
 571                 handle_derefed_pointers(expr->left);
 572                 return true;
 573         }
 574         return false;
 575 }
 576
 577 static void match_assign_host(struct expression *expr)
 578 {
 579         struct symbol *left_type, *right_type;
 580         struct range_list *rl;
 581         static struct expression *handled;
 582         struct smatch_state *state;
 583         struct expression *faked;
 584         bool is_capped = false;
 585         bool is_new = false;
 586
 587         if (!expr)
 588                 return;
 589
 590         left_type = get_type(expr->left);
 591         if (left_type == &void_ctype)
 592                 return;
 593
 594         faked = get_faked_expression();
 595
 596         /* FIXME: handle fake array assignments frob(&host_array[x]); */
 597
 598         if (is_fake_call(expr->right) && faked &&
 599             faked->type == EXPR_ASSIGNMENT &&
 600             points_to_host_data(faked->right)) {
 601                 rl = alloc_whole_rl(get_type(expr->left));
 602                 is_new = true;
 603                 goto set;
 604         }
 605
 606         if (faked && faked == handled)
 607                 return;
 608         if (is_fake_call(expr->right))
 609                 goto clear_old_state;
 610         if (points_to_host_data(expr->right) &&
 611             is_struct_ptr(get_type(expr->left))) {
 612                 handled = expr;
 613                 // This should be handled by smatch_points_to_host_data.c
 614                 //set_points_to_host_data(expr->left);
 615         }
 616
 617         if (handle_op_assign(expr))
 618                 return;
 619         if (expr->op != '=')
 620                 goto clear_old_state;
 621
 622         /* Handled by DB code */
 623         if (expr->right->type == EXPR_CALL)
 624                 return;
 625
 626         if (!get_host_rl(expr->right, &rl))
 627                 goto clear_old_state;
 628
 629         is_capped = host_rl_capped(expr->right);
 630         is_new = state_is_new(expr->right);
 631
 632 set:
 633         if (type_is_ptr(left_type)) {
 634                 right_type = get_type(expr->right);
 635                 if (right_type && right_type->type == SYM_ARRAY)
 636                         set_points_to_host_data(expr->left);
 637                 return;
 638         }
 639
 640         rl = cast_rl(left_type, rl);
 641         state = alloc_estate_rl(rl);
 642         if (is_new)
 643                 estate_set_new(state);
 644         if (is_capped)
 645                 estate_set_capped(state);
 646         set_host_data(expr->left, state);
 647         handle_derefed_pointers(expr->left);
 648         return;
 649
 650 clear_old_state:
 651
 652         /*
 653          * HACK ALERT!!!  This should be at the start of the function.  The
 654          * the problem is that handling "pointer = array;" assignments is
 655          * handled in this function instead of in kernel_points_to_host_data.c.
 656          */
 657         if (type_is_ptr(left_type))
 658                 return;
 659
 660         if (get_state_expr(my_id, expr->left))
 661                 set_host_data(expr->left, alloc_estate_empty());
 662 }
 663
 664 static void handle_eq_noteq(struct expression *expr)
 665 {
 666         struct smatch_state *left_orig, *right_orig;
 667
 668         left_orig = get_state_expr(my_id, expr->left);
 669         right_orig = get_state_expr(my_id, expr->right);
 670
 671         if (!left_orig && !right_orig)
 672                 return;
 673         if (left_orig && right_orig)
 674                 return;
 675
 676         if (left_orig) {
 677                 set_true_false_states_expr(my_id, expr->left,
 678                                 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
 679                                 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
 680         } else {
 681                 set_true_false_states_expr(my_id, expr->right,
 682                                 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
 683                                 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
 684         }
 685 }
 686
 687 static struct range_list *strip_negatives(struct range_list *rl)
 688 {
 689         sval_t min = rl_min(rl);
 690         sval_t minus_one = { .type = rl_type(rl), .value = -1 };
 691         sval_t over = { .type = rl_type(rl), .value = INT_MAX + 1ULL };
 692         sval_t max = sval_type_max(rl_type(rl));
 693
 694         if (!rl)
 695                 return NULL;
 696
 697         if (type_unsigned(rl_type(rl)) && type_bits(rl_type(rl)) > 31)
 698                 return remove_range(rl, over, max);
 699
 700         return remove_range(rl, min, minus_one);
 701 }
 702
 703 static void handle_compare(struct expression *expr)
 704 {
 705         struct expression  *left, *right;
 706         struct range_list *left_rl = NULL;
 707         struct range_list *right_rl = NULL;
 708         struct range_list *host_rl;
 709         struct smatch_state *capped_state;
 710         struct smatch_state *left_true = NULL;
 711         struct smatch_state *left_false = NULL;
 712         struct smatch_state *right_true = NULL;
 713         struct smatch_state *right_false = NULL;
 714         struct symbol *type;
 715         sval_t sval;
 716
 717         left = strip_expr(expr->left);
 718         right = strip_expr(expr->right);
 719
 720         while (left->type == EXPR_ASSIGNMENT)
 721                 left = strip_expr(left->left);
 722
 723         /*
 724          * Conditions are mostly handled by smatch_extra.c, but there are some
 725          * times where the exact values are not known so we can't do that.
 726          *
 727          * Normally, we might consider using smatch_capped.c to supliment smatch
 728          * extra but that doesn't work when we merge unknown uncapped kernel
 729          * data with unknown capped host data.  The result is uncapped host
 730          * data.  We need to keep it separate and say that the host data is
 731          * capped.  In the past, I would have marked this as just regular
 732          * kernel data (not host data) but we can't do that these days because
 733          * we need to track host data for Spectre.
 734          *
 735          * The other situation which we have to handle is when we do have an
 736          * int and we compare against an unknown unsigned kernel variable.  In
 737          * that situation we assume that the kernel data is less than INT_MAX.
 738          * Otherwise then we get all sorts of array underflow false positives.
 739          *
 740          */
 741
 742         /* Handled in smatch_extra.c */
 743         if (get_implied_value(left, &sval) ||
 744             get_implied_value(right, &sval))
 745                 return;
 746
 747         get_host_rl(left, &left_rl);
 748         get_host_rl(right, &right_rl);
 749
 750         /* nothing to do */
 751         if (!left_rl && !right_rl)
 752                 return;
 753         /* if both sides are host data that's not a good limit */
 754         if (left_rl && right_rl)
 755                 return;
 756
 757         if (left_rl)
 758                 host_rl = left_rl;
 759         else
 760                 host_rl = right_rl;
 761
 762         type = get_type(expr);
 763         if (type_unsigned(type))
 764                 host_rl = strip_negatives(host_rl);
 765         capped_state = alloc_estate_rl(host_rl);
 766         estate_set_capped(capped_state);
 767
 768         switch (expr->op) {
 769         case '<':
 770         case SPECIAL_UNSIGNED_LT:
 771         case SPECIAL_LTE:
 772         case SPECIAL_UNSIGNED_LTE:
 773                 if (left_rl)
 774                         left_true = capped_state;
 775                 else
 776                         right_false = capped_state;
 777                 break;
 778         case '>':
 779         case SPECIAL_UNSIGNED_GT:
 780         case SPECIAL_GTE:
 781         case SPECIAL_UNSIGNED_GTE:
 782                 if (left_rl)
 783                         left_false = capped_state;
 784                 else
 785                         right_true = capped_state;
 786                 break;
 787         }
 788         set_true_false_states_expr(my_id, left, left_true, left_false);
 789         set_true_false_states_expr(my_id, right, right_true, right_false);
 790 }
 791
 792 static void match_condition_host(struct expression *expr)
 793 {
 794         if (!expr)
 795                 return;
 796
 797         if (expr->type != EXPR_COMPARE)
 798                 return;
 799
 800         if (expr->op == SPECIAL_EQUAL ||
 801             expr->op == SPECIAL_NOTEQUAL) {
 802                 handle_eq_noteq(expr);
 803                 return;
 804         }
 805         handle_compare(expr);
 806 }
 807
 808 static void match_returns_host_rl(const char *fn, struct expression *expr, void *unused)
 809 {
 810         func_gets_host_data = true;
 811 }
 812
 813 static int has_host_data(struct symbol *sym)
 814 {
 815         struct sm_state *tmp;
 816
 817         FOR_EACH_MY_SM(my_id, __get_cur_stree(), tmp) {
 818                 if (tmp->sym == sym)
 819                         return 1;
 820         } END_FOR_EACH_SM(tmp);
 821         return 0;
 822 }
 823
 824 bool we_pass_host_data(struct expression *call)
 825 {
 826         struct expression *arg;
 827         struct symbol *sym;
 828
 829         FOR_EACH_PTR(call->args, arg) {
 830                 if (points_to_host_data(arg))
 831                         return true;
 832                 sym = expr_to_sym(arg);
 833                 if (!sym)
 834                         continue;
 835                 if (has_host_data(sym))
 836                         return true;
 837         } END_FOR_EACH_PTR(arg);
 838
 839         return false;
 840 }
 841
 842 static int db_returned_host_rl(struct expression *call, struct range_list **rl)
 843 {
 844         struct smatch_state *state;
 845         char buf[48];
 846
 847         if (is_fake_call(call))
 848                 return 0;
 849         snprintf(buf, sizeof(buf), "return %p", call);
 850         state = get_state(my_id, buf, NULL);
 851         if (!state || !estate_rl(state))
 852                 return 0;
 853         *rl = estate_rl(state);
 854         return 1;
 855 }
 856
 857 struct stree *get_host_stree(void)
 858 {
 859         return get_all_states_stree(my_id);
 860 }
 861
 862 static int host_data_flag;
 863 static int no_host_data_flag;
 864
 865 struct range_list *var_host_rl(struct expression *expr)
 866 {
 867         struct smatch_state *state;
 868         struct range_list *rl;
 869         struct range_list *absolute_rl;
 870
 871         if (expr->type == EXPR_PREOP && expr->op == '&') {
 872                 no_host_data_flag = 1;
 873                 return NULL;
 874         }
 875
 876         if (expr->type == EXPR_BINOP && expr->op == '%') {
 877                 struct range_list *left, *right;
 878
 879                 if (!get_host_rl(expr->right, &right))
 880                         return NULL;
 881                 get_absolute_rl(expr->left, &left);
 882                 rl = rl_binop(left, '%', right);
 883                 goto found;
 884         }
 885
 886         if (expr->type == EXPR_BINOP && expr->op == '/') {
 887                 struct range_list *left = NULL;
 888                 struct range_list *right = NULL;
 889                 struct range_list *abs_right;
 890
 891                 get_host_rl(expr->left, &left);
 892                 get_host_rl(expr->right, &right);
 893                 get_absolute_rl(expr->right, &abs_right);
 894
 895                 if (left && !right) {
 896                         rl = rl_binop(left, '/', abs_right);
 897                         if (sval_cmp(rl_max(left), rl_max(rl)) < 0)
 898                                 no_host_data_flag = 1;
 899                 }
 900
 901                 return NULL;
 902         }
 903
 904         if (get_rl_from_function(expr, &rl))
 905                 goto found;
 906
 907         state = get_state_expr(my_id, expr);
 908         if (state && estate_rl(state)) {
 909                 rl = estate_rl(state);
 910                 goto found;
 911         }
 912
 913         if (expr->type == EXPR_CALL && db_returned_host_rl(expr, &rl))
 914                 goto found;
 915
 916         if (expr->type == EXPR_PREOP && expr->op == '*' &&
 917             points_to_host_data(expr->unop)) {
 918                 rl = var_to_absolute_rl(expr);
 919                 goto found;
 920         }
 921
 922         if (is_array(expr)) {
 923                 struct expression *array = get_array_base(expr);
 924
 925                 if (!get_state_expr(my_id, array)) {
 926                         no_host_data_flag = 1;
 927                         return NULL;
 928                 }
 929         }
 930
 931         return NULL;
 932 found:
 933         host_data_flag = 1;
 934         absolute_rl = var_to_absolute_rl(expr);
 935         return clone_rl(rl_intersection(rl, absolute_rl));
 936 }
 937
 938 static bool is_ptr_subtract(struct expression *expr)
 939 {
 940         expr = strip_expr(expr);
 941         if (!expr)
 942                 return false;
 943         if (expr->type == EXPR_BINOP && expr->op == '-' &&
 944             type_is_ptr(get_type(expr->left))) {
 945                 return true;
 946         }
 947         return false;
 948 }
 949
 950 int get_host_rl(struct expression *expr, struct range_list **rl)
 951 {
 952         if (is_ptr_subtract(expr))
 953                 return 0;
 954
 955         host_data_flag = 0;
 956         no_host_data_flag = 0;
 957         custom_get_absolute_rl(expr, &var_host_rl, rl);
 958         if (!host_data_flag || no_host_data_flag)
 959                 *rl = NULL;
 960         return !!*rl;
 961 }
 962
 963 int is_host_rl(struct expression *expr)
 964 {
 965         struct range_list *tmp;
 966
 967         return get_host_rl(expr, &tmp) && tmp;
 968 }
 969
 970 int get_host_rl_var_sym(const char *name, struct symbol *sym, struct range_list **rl)
 971 {
 972         struct smatch_state *state;
 973
 974         state = get_state(my_id, name, sym);
 975         if (state && estate_rl(state)) {
 976                 *rl = estate_rl(state);
 977                 return 1;
 978         }
 979         return 0;
 980 }
 981
 982 static void return_info_callback_host(int return_id, char *return_ranges,
 983                                  struct expression *returned_expr,
 984                                  int param,
 985                                  const char *printed_name,
 986                                  struct sm_state *sm)
 987 {
 988         struct smatch_state *extra;
 989         struct range_list *rl;
 990         char buf[64];
 991
 992         if (param >= 0) {
 993                 if (strcmp(printed_name, "$") == 0)
 994                         return;
 995                 if (!param_was_set_var_sym(sm->name, sm->sym))
 996                         return;
 997         }
 998         rl = estate_rl(sm->state);
 999         if (!rl)
1000                 return;
1001         extra = get_state(SMATCH_EXTRA, sm->name, sm->sym);
1002         if (estate_rl(extra))
1003                 rl = rl_intersection(estate_rl(sm->state), estate_rl(extra));
1004         if (!rl)
1005                 return;
1006
1007         snprintf(buf, sizeof(buf), "%s%s%s",
1008                  show_rl(rl),
1009                  estate_capped(sm->state) ? "[c]" : "", "");
1010         sql_insert_return_states(return_id, return_ranges,
1011                                  estate_new(sm->state) ? HOST_DATA_SET : HOST_DATA,
1012                                  param, printed_name, buf);
1013
1014 }
1015
1016 static void caller_info_callback_host(struct expression *call, int param, char *printed_name, struct sm_state *sm)
1017 {
1018         struct smatch_state *state;
1019         struct range_list *rl;
1020         struct symbol *type;
1021         char buf[64];
1022
1023         /*
1024          * Smatch uses a hack where if we get an unsigned long we say it's
1025          * both host data and it points to host data.  But if we pass it to a
1026          * function which takes an int, then it's just host data.  There's not
1027          * enough bytes for it to be a pointer.
1028          *
1029          */
1030         type = get_arg_type(call->fn, param);
1031         if (strcmp(printed_name, "$") != 0 && type && type_bits(type) < type_bits(&ptr_ctype))
1032                 return;
1033
1034         if (strcmp(sm->state->name, "") == 0)
1035                 return;
1036
1037         state = __get_state(SMATCH_EXTRA, sm->name, sm->sym);
1038         if (!state || !estate_rl(state))
1039                 rl = estate_rl(sm->state);
1040         else
1041                 rl = rl_intersection(estate_rl(sm->state), estate_rl(state));
1042
1043         if (!rl)
1044                 return;
1045
1046         snprintf(buf, sizeof(buf), "%s%s%s", show_rl(rl),
1047                  estate_capped(sm->state) ? "[c]" : "", "");
1048         sql_insert_caller_info(call, HOST_DATA, param, printed_name, buf);
1049 }
1050
1051 static void db_param_set(struct expression *expr, int param, char *key, char *value)
1052 {
1053         struct expression *arg;
1054         char *name;
1055         struct symbol *sym;
1056         struct smatch_state *state;
1057
1058         while (expr->type == EXPR_ASSIGNMENT)
1059                 expr = strip_expr(expr->right);
1060         if (expr->type != EXPR_CALL)
1061                 return;
1062
1063         arg = get_argument_from_call_expr(expr->args, param);
1064         if (!arg)
1065                 return;
1066
1067         name = get_variable_from_key(arg, key, &sym);
1068         if (!name || !sym)
1069                 goto free;
1070
1071         state = get_state(my_id, name, sym);
1072         if (!state)
1073                 goto free;
1074
1075         set_state(my_id, name, sym, alloc_estate_empty());
1076 free:
1077         free_string(name);
1078 }
1079
1080 static bool param_data_capped(const char *value)
1081 {
1082         if (strstr(value, ",c") || strstr(value, "[c"))
1083                 return true;
1084         return false;
1085 }
1086
1087 static void set_param_host_data(const char *name, struct symbol *sym, char *key, char *value)
1088 {
1089         struct expression *expr;
1090         struct range_list *rl = NULL;
1091         struct smatch_state *state;
1092         struct symbol *type;
1093         char *fullname;
1094
1095         expr = symbol_expression(sym);
1096         fullname = get_variable_from_key(expr, key, NULL);
1097         if (!fullname)
1098                 return;
1099
1100         type = get_member_type_from_key(expr, key);
1101         if (type && type->type == SYM_STRUCT)
1102                 return;
1103
1104         if (!type)
1105                 return;
1106
1107         str_to_rl(type, value, &rl);
1108         rl = swap_mtag_seed(expr, rl);
1109         state = alloc_estate_rl(rl);
1110         if (param_data_capped(value) || is_capped(expr))
1111                 estate_set_capped(state);
1112         set_state(my_id, fullname, sym, state);
1113 }
1114
1115 #define OLD 0
1116 #define NEW 1
1117
1118 static void store_host_data_return(struct expression *expr, char *key, char *value, bool is_new)
1119 {
1120         struct smatch_state *state;
1121         struct range_list *rl;
1122         struct symbol *type;
1123         char buf[48];
1124
1125         if (key[0] != '$')
1126                 return;
1127
1128         type = get_type(expr);
1129         snprintf(buf, sizeof(buf), "return %p%s", expr, key + 1);
1130         call_results_to_rl(expr, type, value, &rl);
1131
1132         state = alloc_estate_rl(rl);
1133         if (is_new)
1134                 estate_set_new(state);
1135         set_state(my_id, buf, NULL, state);
1136 }
1137
1138 static void set_to_host_data(struct expression *expr, char *key, char *value, bool is_new)
1139 {
1140         struct smatch_state *state;
1141         char *name;
1142         struct symbol *sym;
1143         struct symbol *type;
1144         struct range_list *rl = NULL;
1145
1146         type = get_member_type_from_key(expr, key);
1147         name = get_variable_from_key(expr, key, &sym);
1148         if (!name || !sym)
1149                 goto free;
1150
1151         call_results_to_rl(expr, type, value, &rl);
1152
1153         state = alloc_estate_rl(rl);
1154         if (param_data_capped(value))
1155                 estate_set_capped(state);
1156         if (is_new)
1157                 estate_set_new(state);
1158         set_state(my_id, name, sym, state);
1159 free:
1160         free_string(name);
1161 }
1162
1163 static void returns_param_host_data(struct expression *expr, int param, char *key, char *value)
1164 {
1165         struct expression *arg;
1166         struct expression *call;
1167
1168         call = expr;
1169         while (call->type == EXPR_ASSIGNMENT)
1170                 call = strip_expr(call->right);
1171         if (call->type != EXPR_CALL)
1172                 return;
1173
1174         if (!we_pass_host_data(call))
1175                 return;
1176
1177         if (param == -1) {
1178                 if (expr->type != EXPR_ASSIGNMENT) {
1179                         store_host_data_return(expr, key, value, OLD);
1180                         return;
1181                 }
1182                 set_to_host_data(expr->left, key, value, OLD);
1183                 return;
1184         }
1185
1186         arg = get_argument_from_call_expr(call->args, param);
1187         if (!arg)
1188                 return;
1189         set_to_host_data(arg, key, value, OLD);
1190 }
1191
1192 static void returns_param_host_data_set(struct expression *expr, int param, char *key, char *value)
1193 {
1194         struct expression *arg;
1195
1196         func_gets_host_data = true;
1197
1198         if (param == -1) {
1199                 if (expr->type != EXPR_ASSIGNMENT) {
1200                         store_host_data_return(expr, key, value, NEW);
1201                         return;
1202                 }
1203                 set_to_host_data(expr->left, key, value, NEW);
1204                 return;
1205         }
1206
1207         while (expr->type == EXPR_ASSIGNMENT)
1208                 expr = strip_expr(expr->right);
1209         if (expr->type != EXPR_CALL)
1210                 return;
1211
1212         arg = get_argument_from_call_expr(expr->args, param);
1213         if (!arg)
1214                 return;
1215         set_to_host_data(arg, key, value, NEW);
1216 }
1217
1218 static void returns_param_capped_host(struct expression *expr, int param, char *key, char *value)
1219 {
1220         struct smatch_state *state, *new;
1221         struct symbol *sym;
1222         char *name;
1223
1224         name = get_name_sym_from_param_key(expr, param, key, &sym);
1225         if (!name || !sym)
1226                 goto free;
1227
1228         state = get_state(my_id, name, sym);
1229         if (!state || estate_capped(state))
1230                 goto free;
1231
1232         new = clone_estate(state);
1233         estate_set_capped(new);
1234
1235         set_state(my_id, name, sym, new);
1236 free:
1237         free_string(name);
1238 }
1239
1240 struct param_key_data {
1241         param_key_hook *call_back;
1242         int param;
1243         const char *key;
1244         void *info;
1245 };
1246
1247 static void match_function_def(struct symbol *sym)
1248 {
1249         if (is_host_data_fn(sym))
1250                 func_gets_host_data = true;
1251 }
1252
1253 static void set_param_host_input_data(struct expression *expr, const char *name,
1254                                  struct symbol *sym, void *data)
1255 {
1256     struct expression *arg;
1257
1258     func_gets_host_data = true;
1259         arg = gen_expression_from_name_sym(name, sym);
1260         tag_as_host_data(arg);
1261     return;
1262 }
1263
1264 void register_kernel_host_data(int id)
1265 {
1266         int i;
1267         struct host_fn_info *info;
1268
1269         my_id = id;
1270
1271         if (option_project != PROJ_KERNEL)
1272                 return;
1273
1274         set_dynamic_states(my_id);
1275         add_function_data(&func_gets_host_data);
1276         add_hook(&match_function_def, FUNC_DEF_HOOK);
1277
1278         add_function_data((unsigned long *)&start_states);
1279         add_unmatched_state_hook(my_id, &empty_state);
1280         add_extra_nomod_hook(&extra_nomod_hook);
1281         add_pre_merge_hook(my_id, &pre_merge_hook);
1282         add_merge_hook(my_id, &merge_estates);
1283
1284
1285         for (i = 0; i < ARRAY_SIZE(func_table); i++) {
1286                 info = &func_table[i];
1287                 add_function_param_key_hook_late(info->name, &set_param_host_input_data,
1288                                                                                  info->param, info->key, info);
1289         }
1290
1291         for (i = 0; i < ARRAY_SIZE(func_table); i++) {
1292                 if (func_table[i].param == -1)
1293                         add_function_hook(func_table[i].name, &match_returns_host_rl, NULL);
1294         }
1295
1296         add_hook(&match_assign_host, ASSIGNMENT_HOOK);
1297         select_return_states_hook(PARAM_SET, &db_param_set);
1298         add_hook(&match_condition_host, CONDITION_HOOK);
1299
1300         add_caller_info_callback(my_id, caller_info_callback_host);
1301         add_return_info_callback(my_id, return_info_callback_host);
1302         select_caller_info_hook(set_param_host_data, HOST_DATA);
1303         select_return_states_hook(HOST_DATA, &returns_param_host_data);
1304         select_return_states_hook(HOST_DATA_SET, &returns_param_host_data_set);
1305         select_return_states_hook(CAPPED_DATA, &returns_param_capped_host);
1306 }