[REVIEW REQUEST] staging: unisys: review request for visorbus
Dan Carpenter
dan.carpenter at oracle.com
Sat Nov 4 06:48:58 UTC 2017
drivers/staging/unisys/visorbus/visorchipset.c
588 static void *parser_name_get(struct parser_context *ctx)
589 {
590 struct visor_controlvm_parameters_header *phdr;
591
592 phdr = &ctx->data;
593 if (phdr->name_offset + phdr->name_length > ctx->param_bytes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I am just reviewing quickly. This looks like a potential integer
overflow if phdr->name_offset is UINT_MAX?
594 return NULL;
595 ctx->curr = (char *)&phdr + phdr->name_offset;
596 ctx->bytes_remaining = phdr->name_length;
597 return parser_string_get(ctx->curr, phdr->name_length);
598 }
drivers/staging/unisys/visorbus/visorchipset.c
1317 static struct parser_context *parser_init_stream(u64 addr, u32 bytes,
1318 bool *retry)
1319 {
1320 int allocbytes;
1321 struct parser_context *ctx;
1322 void *mapping;
1323
1324 *retry = false;
1325 /* alloc an extra byte to ensure payload is \0 terminated */
1326 allocbytes = bytes + 1 + (sizeof(struct parser_context) -
1327 sizeof(struct visor_controlvm_parameters_header));
1328 if ((chipset_dev->controlvm_payload_bytes_buffered + bytes) >
Same question. Can this or the allocbytes calculation have an integer
overflow?
1329 MAX_CONTROLVM_PAYLOAD_BYTES) {
1330 *retry = true;
1331 return NULL;
1332 }
1333 ctx = kzalloc(allocbytes, GFP_KERNEL);
1334 if (!ctx) {
1335 *retry = true;
1336 return NULL;
1337 }
1338 ctx->allocbytes = allocbytes;
1339 ctx->param_bytes = bytes;
Otherwise the code looks pretty clean to me.
regards,
dan carpenter
More information about the devel
mailing list