diff options
author | Bohdan Tymkiv <bhdt@cypress.com> | 2019-04-16 17:28:29 +0300 |
---|---|---|
committer | Tomas Vanek <vanekt@fbl.cz> | 2019-06-04 21:37:41 +0100 |
commit | eea508d9afa184684bf5a4e2788b6cd4b5fdc9ab (patch) | |
tree | b5aaf19b1591bf78d69d90876449b0d528f83d1f | |
parent | 85ed41d2100ba2f451ba8ff3d75abbeb8cf10a29 (diff) | |
download | riscv-openocd-eea508d9afa184684bf5a4e2788b6cd4b5fdc9ab.zip riscv-openocd-eea508d9afa184684bf5a4e2788b6cd4b5fdc9ab.tar.gz riscv-openocd-eea508d9afa184684bf5a4e2788b6cd4b5fdc9ab.tar.bz2 |
gdb_server: fix GDB_BUFFER_SIZE usage, fix unaligned access during bulk transfers
Currently size of the GDB buffer is 16384 bytes but it is treated as
nul-terminated string in most of the code, so effective size of the
buffer is actually 16383 bytes. OpenOCD responds with `PacketSize=3fff`
to qSupported request. Result of GDB's `m` command is encoded in hex so
each data byte uses two bytes in the buffer. As a result GDB will split
bulk read requests into chunks 0x1fff bytes each. This causes troubles
on targets (or memory regions) which support only aligned, word-sized
access (such as MMIO buffers).
Steps to reproduce (psoc6 target):
gdb> dump binary memory dump.bin 0x040320000 (0x040320000 + 65536)
OpenOCD:
Error: Failed to read memory at 0x40321ffe
Error: Failed to read memory at 0x40321000
Error: Failed to read memory at 0x40323000
Error: Failed to read memory at 0x40325ffe
Error: Failed to read memory at 0x40329ffa
Error: Failed to read memory at 0x40329ffc
Error: Failed to read memory at 0x4032bffc
Error: Failed to read memory at 0x4032dffa
Consolidate GDB_BUFFER_SIZE usage: ensure size of each buffer is
(GDB_BUFFER_SIZE + 1), add explicit comment that additional byte is used
for nul-termination. Report correct size of the buffer to GDB (0x4000)
as recommended in GDB's docummentation: `if the stub stores packets in a
NUL-terminated format, it should allow an extra byte in its buffer for
the NUL`
Checked with clang-asan, clang-analyzer, valgrind - no new errors.
Change-Id: I909e8a2c6b010c5d4a304641808d4a807a4ec18d
Signed-off-by: Bohdan Tymkiv <bhdt@cypress.com>
Reviewed-on: http://openocd.zylin.com/5109
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
-rw-r--r-- | src/rtos/nuttx.c | 2 | ||||
-rw-r--r-- | src/rtos/rtos.c | 2 | ||||
-rw-r--r-- | src/server/gdb_server.c | 10 |
3 files changed, 6 insertions, 8 deletions
diff --git a/src/rtos/nuttx.c b/src/rtos/nuttx.c index 61fd9aa..8c3076e 100644 --- a/src/rtos/nuttx.c +++ b/src/rtos/nuttx.c @@ -174,7 +174,7 @@ static int rcmd_offset(const char *cmd, const char *name) static int nuttx_thread_packet(struct connection *connection, char const *packet, int packet_size) { - char cmd[GDB_BUFFER_SIZE / 2] = ""; + char cmd[GDB_BUFFER_SIZE / 2 + 1] = ""; /* Extra byte for nul-termination */ if (!strncmp(packet, "qRcmd", 5)) { size_t len = unhexify((uint8_t *)cmd, packet + 6, sizeof(cmd)); diff --git a/src/rtos/rtos.c b/src/rtos/rtos.c index da0a503..20e875d 100644 --- a/src/rtos/rtos.c +++ b/src/rtos/rtos.c @@ -223,7 +223,7 @@ int rtos_qsymbol(struct connection *connection, char const *packet, int packet_s int rtos_detected = 0; uint64_t addr = 0; size_t reply_len; - char reply[GDB_BUFFER_SIZE], cur_sym[GDB_BUFFER_SIZE / 2] = ""; + char reply[GDB_BUFFER_SIZE + 1], cur_sym[GDB_BUFFER_SIZE / 2 + 1] = ""; /* Extra byte for nul-termination */ symbol_table_elem_t *next_sym = NULL; struct target *target = get_target_from_connection(connection); struct rtos *os = target->rtos; diff --git a/src/server/gdb_server.c b/src/server/gdb_server.c index c6cf7d3..60ed4ce 100644 --- a/src/server/gdb_server.c +++ b/src/server/gdb_server.c @@ -67,7 +67,7 @@ struct target_desc_format { /* private connection data for GDB */ struct gdb_connection { - char buffer[GDB_BUFFER_SIZE]; + char buffer[GDB_BUFFER_SIZE + 1]; /* Extra byte for nul-termination */ char *buf_p; int buf_cnt; int ctrl_c; @@ -1407,8 +1407,6 @@ static int gdb_error(struct connection *connection, int retval) /* We don't have to worry about the default 2 second timeout for GDB packets, * because GDB breaks up large memory reads into smaller reads. - * - * 8191 bytes by the looks of it. Why 8191 bytes instead of 8192????? */ static int gdb_read_memory_packet(struct connection *connection, char const *packet, int packet_size) @@ -2614,7 +2612,7 @@ static int gdb_query_packet(struct connection *connection, &pos, &size, "PacketSize=%x;qXfer:memory-map:read%c;qXfer:features:read%c;qXfer:threads:read+;QStartNoAckMode+;vContSupported+", - (GDB_BUFFER_SIZE - 1), + GDB_BUFFER_SIZE, ((gdb_use_memory_map == 1) && (flash_get_bank_count() > 0)) ? '+' : '-', (gdb_target_desc_supported == 1) ? '+' : '-'); @@ -3117,7 +3115,7 @@ static void gdb_sig_halted(struct connection *connection) static int gdb_input_inner(struct connection *connection) { /* Do not allocate this on the stack */ - static char gdb_packet_buffer[GDB_BUFFER_SIZE]; + static char gdb_packet_buffer[GDB_BUFFER_SIZE + 1]; /* Extra byte for nul-termination */ struct target *target; char const *packet = gdb_packet_buffer; @@ -3140,7 +3138,7 @@ static int gdb_input_inner(struct connection *connection) * drain the rest of the buffer. */ do { - packet_size = GDB_BUFFER_SIZE-1; + packet_size = GDB_BUFFER_SIZE; retval = gdb_get_packet(connection, gdb_packet_buffer, &packet_size); if (retval != ERROR_OK) return retval; |