From mboxrd@z Thu Jan 1 00:00:00 1970 From: Oleksandr Tymoshenko Date: Thu, 26 Jan 2017 11:50:19 -0800 Subject: [U-Boot] [PATCH] fs/fat: Fix unaligned __u16 reads for FAT12 access In-Reply-To: <2b77133c2d5a409793fcd8d4764b40fc@rwthex-w2-b.rwth-ad.de> References: <20170125232546.GA93084@bluezbox.com> <2b77133c2d5a409793fcd8d4764b40fc@rwthex-w2-b.rwth-ad.de> Message-ID: <20170126195019.GA398@bluezbox.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Br?ns, Stefan (Stefan.Bruens at rwth-aachen.de) wrote: > Doing unaligned reads is not supported on all architectures, use > byte sized reads of the little endian buffer. > Rename off16 to off8, as it reflects the buffer offset in byte > granularity (offset is in entry, i.e. 12 bit, granularity). > Fix a regression introduced in 8d48c92b45aea91e2a2be90f2ed93677e85526f1 > > Reported-by: Oleksandr Tymoshenko > Signed-off-by: Stefan Br?ns > --- > fs/fat/fat.c | 7 ++++--- > 1 file changed, 4 insertions(+), 3 deletions(-) > > diff --git a/fs/fat/fat.c b/fs/fat/fat.c > index fe899d0442..06088e2421 100644 > --- a/fs/fat/fat.c > +++ b/fs/fat/fat.c > @@ -179,7 +179,7 @@ int flush_dirty_fat_buffer(fsdata *mydata) > static __u32 get_fatent(fsdata *mydata, __u32 entry) > { > __u32 bufnum; > - __u32 off16, offset; > + __u32 offset, off8; > __u32 ret = 0x00; > > if (CHECK_CLUST(entry, mydata->fatsize)) { > @@ -242,8 +242,9 @@ static __u32 get_fatent(fsdata *mydata, __u32 entry) > ret = FAT2CPU16(((__u16 *) mydata->fatbuf)[offset]); > break; > case 12: > - off16 = (offset * 3) / 2; > - ret = FAT2CPU16(*(__u16 *)(mydata->fatbuf + off16)); > + off8 = (offset * 3) / 2; > + /* fatbut + off8 may be unaligned, read in byte granularity */ > + ret = mydata->fatbuf[off8] + mydata->fatbuf[off8 + 1] << 8; This patch when applied gives me "Invalid FAT entry" message. Reason is: operator '<<' has lower precedence than '+' so this expression is equivalent to (mydata->fatbuf[off8] + mydata->fatbuf[off8 + 1]) << 8 With explicit parentheses around shift it works as expected: ret = mydata->fatbuf[off8] + (mydata->fatbuf[off8 + 1] << 8); -- gonzo