| Message ID | h077d2$ra9$1@ger.gmane.org |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, 2009-06-03 at 18:18 -0700, Nahor wrote: > --- a/nandwrite.c 2009-06-01 18:03:55.000000000 -0700 > +++ b/nandwrite.c 2009-06-03 15:29:24.000000000 -0700 > @@ -531,7 +531,7 @@ > if (writeoob) { > int tinycnt = 0; > > - while(tinycnt < readlen) { > + while(tinycnt < meminfo.oobsize) { > cnt = read(ifd, oobreadbuf + tinycnt, > meminfo.oobsize - tinycnt); > if (cnt == 0) { // EOF > break; > This looks valid for me. Would you please, send a patch which I can apply, which would include your name and Signed-off-by signature?
--- a/nandwrite.c 2009-06-01 18:03:55.000000000 -0700 +++ b/nandwrite.c 2009-06-03 15:29:24.000000000 -0700 @@ -531,7 +531,7 @@ if (writeoob) { int tinycnt = 0; - while(tinycnt < readlen) { + while(tinycnt < meminfo.oobsize) { cnt = read(ifd, oobreadbuf + tinycnt, meminfo.oobsize - tinycnt); if (cnt == 0) { // EOF
Hi all, Here is a small patch to fix an obvious typo. If the file contains only a few bytes in the last page and the padding option is selected, the loop may exit early (readlen < meminfo.oobsize). Most of the time it will still work though because the code tries to read the whole OOB in one chunk. On a similar issue, shouldn't the loop reading the input file exit with error if cnt == 0? The loop should alread exit when all the expected data is read. So if cnt == 0, it means that nandwrite didn't get all the data it wanted. The code then goes on as if everything was correctly read so it will write the data from the previous page. This is a security risk (even though the necessary conditions make it nearly impossible in practice). Regards, Nahor break;