Woops, forgot about the Signed-off-by. Sorry!
On 9/14/20 5:51 PM, Ahmad Fatoum wrote:
Hello Yann,
On 9/14/20 5:27 PM, Yann Sionneau wrote:
char type can be either signed or unsigned according to C standard.
If your arch has signed char, this kind of computation will end up wrong
because of sign extension:
https://git.pengutronix.de/cgit/barebox/tree/drivers/mci/mci-core.c#n869
mci->capacity = mci->ext_csd[EXT_CSD_SEC_COUNT] << 0 |
mci->ext_csd[EXT_CSD_SEC_COUNT + 1] << 8 |
mci->ext_csd[EXT_CSD_SEC_COUNT + 2] << 16 |
mci->ext_csd[EXT_CSD_SEC_COUNT + 3] << 24;
Good catch! I wondered why this wasn't noticed before,
so I looked at the OSELAS.Toolchain`s we have on our build
server. Only x86 and MIPS have signed chars by default.
On x86, we usually do EFI, so barebox mci-core is not involved
and MIPS apparently doesn't have any MCI drivers yet.
(Of course, you can compile do -fsigned-char or compile
your own toolchain with a different default, but I guess
not many do that)
Thanks for fixing it.
Turning the ext_csd field into u8 * fixes the issue.
Your Signed-off-by is required.
Please check https://developercertificate.org/
---
include/mci.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/mci.h b/include/mci.h
index 96547fb39..f6d845440 100644
--- a/include/mci.h
+++ b/include/mci.h
@@ -456,7 +456,7 @@ struct mci {
uint64_t capacity; /**< Card's data capacity in bytes */
int ready_for_use; /** true if already probed */
int dsr_imp; /**< DSR implementation state from CSD */
- char *ext_csd;
+ u8 *ext_csd;
int probe;
struct param_d *param_boot;
int bootpart;
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
