void tbranch_on_branch_taint2(Addr a, uint64_t size) {
if (pandalog) {
// a is an llvm reg
assert (a.typ == LADDR);
// count number of tainted bytes on this reg
// NB: assuming 8 bytes
uint32_t num_tainted = 0;
for (uint32_t o=0; o<size; o++) {
Addr ao =a;
ao.off = o;
num_tainted += (taint2_query(ao) != 0);
}
if (num_tainted > 0) {
if (summary) {
CPUState *cpu = first_cpu;
target_ulong asid = panda_current_asid(cpu);
tainted_branch[asid].insert(panda_current_pc(cpu));
}
else {
Panda__TaintedBranch *tb = (Panda__TaintedBranch *) malloc(sizeof(Panda__TaintedBranch));
*tb = PANDA__TAINTED_BRANCH__INIT;
tb->call_stack = pandalog_callstack_create();
tb->n_taint_query = num_tainted;
tb->taint_query = (Panda__TaintQuery **) malloc (sizeof (Panda__TaintQuery *) * num_tainted);
uint32_t i=0;
for (uint32_t o=0; o<size; o++) {
Addr ao = a;
ao.off = o;
if (taint2_query(ao)) {
tb->taint_query[i++] = taint2_query_pandalog(ao, o);
}
}
Panda__LogEntry ple = PANDA__LOG_ENTRY__INIT;
ple.tainted_branch = tb;
pandalog_write_entry(&ple);
pandalog_callstack_free(tb->call_stack);
for (uint32_t i=0; i<num_tainted; i++) {
pandalog_taint_query_free(tb->taint_query[i]);
}
free(tb);
}
}
}
}
// hypercall-initiated taint query of some src-level extent
void lava_taint_query (Panda__SrcInfoPri *si, target_ulong buf, target_ulong buf_len) {
extern CPUState *cpu_single_env;
CPUState *env = cpu_single_env;
//if (pandalog && taintEnabled && (taint2_num_labels_applied() > 0)){
if (pandalog && taint2_enabled() && (taint2_num_labels_applied() > 0)){
// okay, taint is on and some labels have actually been applied
// is there *any* taint on this extent
uint32_t num_tainted = 0;
bool is_strnlen = false;
//bool is_strnlen = ((int) phs.len == -1);
uint32_t offset=0;
while (true) {
// for (uint32_t offset=0; offset<phs.len; offset++) {
uint32_t va = buf + offset;
//uint32_t va = phs.buf + offset;
uint32_t pa = panda_virt_to_phys(env, va);
if (is_strnlen) {
uint8_t c;
panda_virtual_memory_rw(env, pa, &c, 1, false);
// null terminator
if (c==0) break;
}
if ((int) pa != -1) {
Addr a = make_maddr(pa);
if (taint2_query(a)) {
num_tainted ++;
}
}
offset ++;
// end of query by length or max string length
if (!is_strnlen && offset == buf_len) break;
//if (!is_strnlen && offset == phs.len) break;
if (is_strnlen && (offset == LAVA_TAINT_QUERY_MAX_LEN)) break;
}
uint32_t len = offset;
if (num_tainted) {
printf("logging lava query\n");
// ok at least one byte in the extent is tainted
// 1. write the pandalog entry that tells us something was tainted on this extent
Panda__TaintQueryPri *tqh = (Panda__TaintQueryPri *) malloc (sizeof (Panda__TaintQueryPri));
*tqh = PANDA__TAINT_QUERY_PRI__INIT;
tqh->buf = buf;
//tqh->buf = phs.buf;
tqh->len = len;
tqh->num_tainted = num_tainted;
// obtain the actual data out of memory
// NOTE: first X bytes only!
uint32_t data[LAVA_TAINT_QUERY_MAX_LEN];
uint32_t n = len;
// grab at most X bytes from memory to pandalog
// this is just a snippet. we dont want to write 1M buffer
if (LAVA_TAINT_QUERY_MAX_LEN < len) n = LAVA_TAINT_QUERY_MAX_LEN;
for (uint32_t i=0; i<n; i++) {
data[i] = 0;
uint8_t c;
panda_virtual_memory_rw(env, buf+i, &c, 1, false);
//panda_virtual_memory_rw(env, phs.buf+i, &c, 1, false);
data[i] = c;
}
tqh->n_data = n;
tqh->data = data;
// 2. write out src-level info
//Panda__SrcInfoPri *si = pandalog_src_info_create(phs);
tqh->src_info = si;
// 3. write out callstack info
Panda__CallStack *cs = pandalog_callstack_create();
tqh->call_stack = cs;
// 4. iterate over the bytes in the extent and pandalog detailed info about taint
std::vector<Panda__TaintQuery *> tq;
for (uint32_t offset=0; offset<len; offset++) {
uint32_t va = buf + offset;
//uint32_t va = phs.buf + offset;
uint32_t pa = panda_virt_to_phys(env, va);
if ((int) pa != -1) {
Addr a = make_maddr(pa);
if (taint2_query(a)) {
tq.push_back(taint2_query_pandalog(a, offset));
}
}
}
tqh->n_taint_query = tq.size();
tqh->taint_query = (Panda__TaintQuery **) malloc(sizeof(Panda__TaintQuery *) * tqh->n_taint_query);
for (uint32_t i=0; i<tqh->n_taint_query; i++) {
tqh->taint_query[i] = tq[i];
}
Panda__LogEntry ple = PANDA__LOG_ENTRY__INIT;
ple.taint_query_pri = tqh;
printf("about to write out taint query entry\n");
pandalog_write_entry(&ple);
free(tqh->src_info);
pandalog_callstack_free(tqh->call_stack);
for (uint32_t i=0; i<tqh->n_taint_query; i++) {
pandalog_taint_query_free(tqh->taint_query[i]);
}
free(tqh);
}
}
}
/**
* @brief Hypercall-initiated taint query of some src-level extent.
*/
void taint_query_hypercall(PandaHypercallStruct phs) {
CPUState *cpu = first_cpu;
if (pandalog && taintEnabled && (taint2_num_labels_applied() > 0)) {
// okay, taint is on and some labels have actually been applied
// is there *any* taint on this extent
uint32_t num_tainted = 0;
bool is_strnlen = ((int) phs.len == -1);
uint32_t offset=0;
while (true) {
uint32_t va = phs.buf + offset;
uint32_t pa = panda_virt_to_phys(cpu, va);
if (is_strnlen) {
uint8_t c;
panda_virtual_memory_rw(cpu, pa, &c, 1, false);
// null terminator
if (c==0) break;
}
if ((int) pa != -1) {
Addr a = make_maddr(pa);
if (taint2_query(a)) {
num_tainted ++;
}
}
offset ++;
// end of query by length or max string length
if (!is_strnlen && offset == phs.len) break;
if (is_strnlen && (offset == QUERY_HYPERCALL_MAX_LEN)) break;
}
uint32_t len = offset;
if (num_tainted) {
// ok at least one byte in the extent is tainted
// 1. write the pandalog entry that tells us something was tainted on this extent
Panda__TaintQueryHypercall *tqh = (Panda__TaintQueryHypercall *) malloc (sizeof (Panda__TaintQueryHypercall));
*tqh = PANDA__TAINT_QUERY_HYPERCALL__INIT;
tqh->buf = phs.buf;
tqh->len = len;
tqh->num_tainted = num_tainted;
// obtain the actual data out of memory
// NOTE: first X bytes only!
uint32_t data[QUERY_HYPERCALL_MAX_LEN];
uint32_t n = len;
// grab at most X bytes from memory to pandalog
// this is just a snippet. we dont want to write 1M buffer
if (QUERY_HYPERCALL_MAX_LEN < len) n = QUERY_HYPERCALL_MAX_LEN;
for (uint32_t i=0; i<n; i++) {
data[i] = 0;
uint8_t c;
panda_virtual_memory_rw(cpu, phs.buf+i, &c, 1, false);
data[i] = c;
}
tqh->n_data = n;
tqh->data = data;
// 2. write out src-level info
Panda__SrcInfo *si = pandalog_src_info_create(phs);
tqh->src_info = si;
// 3. write out callstack info
Panda__CallStack *cs = pandalog_callstack_create();
tqh->call_stack = cs;
std::vector<Panda__TaintQuery *> tq;
for (uint32_t offset=0; offset<len; offset++) {
uint32_t va = phs.buf + offset;
uint32_t pa = panda_virt_to_phys(cpu, va);
if ((int) pa != -1) {
Addr a = make_maddr(pa);
if (taint2_query(a)) {
tq.push_back(taint2_query_pandalog(a, offset));
}
}
}
tqh->n_taint_query = tq.size();
tqh->taint_query = (Panda__TaintQuery **) malloc(sizeof(Panda__TaintQuery *) * tqh->n_taint_query);
for (uint32_t i=0; i<tqh->n_taint_query; i++) {
tqh->taint_query[i] = tq[i];
}
Panda__LogEntry ple = PANDA__LOG_ENTRY__INIT;
ple.taint_query_hypercall = tqh;
pandalog_write_entry(&ple);
free(tqh->src_info);
pandalog_callstack_free(tqh->call_stack);
for (uint32_t i=0; i<tqh->n_taint_query; i++) {
pandalog_taint_query_free(tqh->taint_query[i]);
}
free(tqh);
}
}
}