gdb_server: correctly report flash sector sizes

Report each region of same-size sectors separately, instead of
incorrectly reporting that every sector has the same size.

This is a longstanding bug on NOR flash chips with non-uniform
sector sizes.  It was largely hidden by other bugs in flash
handling.  When some of those were recently fixed, this one was
exposed as a regression on str710.

[oyvind.harboe@zylin.com: update the loop to behave on str7 ]

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
This commit is contained in:
David Brownell 2010-01-20 10:43:32 -08:00
parent d036f17001
commit 6f2b88448f
1 changed files with 46 additions and 31 deletions

View File

@ -1613,22 +1613,6 @@ static int decode_xfer_read(char *buf, char **annex, int *ofs, unsigned int *len
return 0;
}
static int gdb_calc_blocksize(struct flash_bank *bank)
{
uint32_t i;
uint32_t block_size = 0xffffffff;
/* loop through all sectors and return smallest sector size */
for (i = 0; i < (uint32_t)bank->num_sectors; i++)
{
if (bank->sectors[i].size < block_size)
block_size = bank->sectors[i].size;
}
return block_size;
}
static int compare_bank (const void * a, const void * b)
{
struct flash_bank *b1, *b2;
@ -1666,7 +1650,6 @@ static int gdb_memory_map(struct connection *connection,
int offset;
int length;
char *separator;
int blocksize;
uint32_t ram_start = 0;
int i;
@ -1683,6 +1666,7 @@ static int gdb_memory_map(struct connection *connection,
* it has no concept of non-cacheable read/write memory (i/o etc).
*
* FIXME Most non-flash addresses are *NOT* RAM! Don't lie.
* Current versions of GDB assume unlisted addresses are RAM...
*/
banks = malloc(sizeof(struct flash_bank *)*flash_get_bank_count());
@ -1701,29 +1685,60 @@ static int gdb_memory_map(struct connection *connection,
compare_bank);
for (i = 0; i < flash_get_bank_count(); i++) {
int j;
unsigned sector_size = 0;
uint32_t start, end;
p = banks[i];
start = p->base;
end = p->base + p->size;
if (ram_start < p->base)
xml_printf(&retval, &xml, &pos, &size,
"<memory type=\"ram\" start=\"0x%x\" "
"length=\"0x%x\"/>\n",
ram_start, p->base-ram_start);
ram_start, p->base - ram_start);
/* If device has uneven sector sizes, eg. str7, lpc
* we pass the smallest sector size to gdb memory map
*
* FIXME Don't lie about flash regions with different
* sector sizes; just tell GDB about each region as
* if it were a separate flash device.
/* Report adjacent groups of same-size sectors. So for
* example top boot CFI flash will list an initial region
* with several large sectors (maybe 128KB) and several
* smaller ones at the end (maybe 32KB). STR7 will have
* regions with 8KB, 32KB, and 64KB sectors; etc.
*/
blocksize = gdb_calc_blocksize(p);
for (j = 0; j < p->num_sectors; j++) {
unsigned group_len;
/* Maybe start a new group of sectors. */
if (sector_size == 0) {
start = p->base + p->sectors[j].offset;
xml_printf(&retval, &xml, &pos, &size,
"<memory type=\"flash\" "
"start=\"0x%x\" ",
start);
sector_size = p->sectors[j].size;
}
/* Does this finish a group of sectors?
* If not, continue an already-started group.
*/
if (j == p->num_sectors -1)
group_len = (p->base + p->size) - start;
else if (p->sectors[j + 1].size != sector_size)
group_len = p->base + p->sectors[j + 1].offset
- start;
else
continue;
xml_printf(&retval, &xml, &pos, &size,
"length=\"0x%x\">\n"
"<property name=\"blocksize\">"
"0x%x</property>\n"
"</memory>\n",
group_len,
sector_size);
sector_size = 0;
}
xml_printf(&retval, &xml, &pos, &size,
"<memory type=\"flash\" start=\"0x%x\" "
"length=\"0x%x\">\n" \
"<property name=\"blocksize\">0x%x</property>\n" \
"</memory>\n", \
p->base, p->size, blocksize);
ram_start = p->base + p->size;
}