So far, writing a barebox image exceeding the partition size aborts with
EPERM as truncate isn't implemented:
ERROR: writing to /dev/flash-boot.barebox failed with Operation
not permitted
update failed
ERROR: fastboot: update barebox: Operation not permitted
This is unfortunate because by the time the truncation fails, erasing
the partition had already occurred. Avoid this by checking prior to the
pwrite_all whether the file to be written is big enough. This is valid
here because barebox update wouldn't be called on a regular file.
While at it, present callers with a more helpful ENOSPC error.
Signed-off-by: Ahmad Fatoum <a.fatoum@xxxxxxxxxxxxxx>
---
arch/arm/mach-imx/imx-bbu-internal.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/arch/arm/mach-imx/imx-bbu-internal.c b/arch/arm/mach-imx/imx-bbu-internal.c
index a563b3bc2906..946a3e9a779b 100644
--- a/arch/arm/mach-imx/imx-bbu-internal.c
+++ b/arch/arm/mach-imx/imx-bbu-internal.c
@@ -87,6 +87,7 @@ static int imx_bbu_write_device(struct imx_internal_bbu_handler *imx_handler,
const void *buf, int image_len)
{
int fd, ret, offset = 0;
+ struct stat st;
fd = open(devicefile, O_RDWR | O_CREAT);
if (fd < 0)
@@ -101,6 +102,15 @@ static int imx_bbu_write_device(struct imx_internal_bbu_handler *imx_handler,
if (imx_handler->handler.flags & IMX_BBU_FLAG_KEEP_HEAD)
offset += imx_handler->flash_header_offset;
+ ret = fstat(fd, &st);
+ if (ret)
+ goto err_close;
+
+ if (image_len > st.st_size) {
+ ret = -ENOSPC;
+ goto err_close;
+ }
+
ret = imx_bbu_protect(fd, imx_handler, devicefile, offset,
image_len, 0);
if (ret)
--
2.20.1
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
