diff mbox

[U-Boot] catch wrong load address passed to fatload / ext2load

Message ID 20140709204257.GA28671@amd.pavel.ucw.cz
State Accepted
Delegated to: Tom Rini
Headers show

Commit Message

Pavel Machek July 9, 2014, 8:42 p.m. UTC
If filename is passed instead of address to ext2load or fatload,
u-boot silently accepts that, and uses 0 for load address and default
filename from environment. That is confusing, display help instead.
    
Signed-off-by: Pavel Machek <pavel@denx.de>

Comments

Pavel Machek July 9, 2014, 8:46 p.m. UTC | #1
Hi!

> If filename is passed instead of address to ext2load or fatload,
> u-boot silently accepts that, and uses 0 for load address and default
> filename from environment. That is confusing, display help instead.
>     
> Signed-off-by: Pavel Machek <pavel@denx.de>

Oops and actually I should warn... this patch is untested. I'm
currently fighting hard to get recent u-boot to run on socfpga.

								Pavel
Marek Vasut July 9, 2014, 9:07 p.m. UTC | #2
On Wednesday, July 09, 2014 at 10:42:57 PM, Pavel Machek wrote:
> If filename is passed instead of address to ext2load or fatload,
> u-boot silently accepts that, and uses 0 for load address and default
> filename from environment. That is confusing, display help instead.
> 
> Signed-off-by: Pavel Machek <pavel@denx.de>

The handling of simple_strtoul() in fs.c could use improvement like this in 
general. Do you mind expanding this patch ?

Best regards,
Marek Vasut
Pavel Machek July 9, 2014, 10:57 p.m. UTC | #3
On Wed 2014-07-09 23:07:56, Marek Vasut wrote:
> On Wednesday, July 09, 2014 at 10:42:57 PM, Pavel Machek wrote:
> > If filename is passed instead of address to ext2load or fatload,
> > u-boot silently accepts that, and uses 0 for load address and default
> > filename from environment. That is confusing, display help instead.
> > 
> > Signed-off-by: Pavel Machek <pavel@denx.de>
> 
> The handling of simple_strtoul() in fs.c could use improvement like this in 
> general. Do you mind expanding this patch ?

I'd prefer not to do that at this moment: I don't have working-enough
u-boot and hate to do changes without testing.

									Pavel
Wolfgang Denk July 9, 2014, 11:08 p.m. UTC | #4
Dear Pavel Machek,

In message <20140709204257.GA28671@amd.pavel.ucw.cz> you wrote:
> 
> If filename is passed instead of address to ext2load or fatload,
> u-boot silently accepts that, and uses 0 for load address and default
> filename from environment. That is confusing, display help instead.
>     
> Signed-off-by: Pavel Machek <pavel@denx.de>
> 
> diff --git a/fs/fs.c b/fs/fs.c
> index 79d432d..ea15c5f 100644
> --- a/fs/fs.c
> +++ b/fs/fs.c
> @@ -276,6 +276,7 @@ int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
>  	unsigned long pos;
>  	int len_read;
>  	unsigned long time;
> +	char *ep;
>  
>  	if (argc < 2)
>  		return CMD_RET_USAGE;
> @@ -286,7 +287,9 @@ int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
>  		return 1;
>  
>  	if (argc >= 4) {
> -		addr = simple_strtoul(argv[3], NULL, 16);
> +		addr = simple_strtoul(argv[3], &ep, 16);
> +		if (ep == argv[3] || *ep != '\0')
> +			return CMD_RET_USAGE;

What happens in case of filenames that look like numbers, say "0"?
It may be silly to use such, but it should not be impossible to do so.

Viele Grüße,

Wolfgang Denk
Pavel Machek July 9, 2014, 11:12 p.m. UTC | #5
On Thu 2014-07-10 01:08:28, Wolfgang Denk wrote:
> Dear Pavel Machek,
> 
> In message <20140709204257.GA28671@amd.pavel.ucw.cz> you wrote:
> > 
> > If filename is passed instead of address to ext2load or fatload,
> > u-boot silently accepts that, and uses 0 for load address and default
> > filename from environment. That is confusing, display help instead.
> >     
> > Signed-off-by: Pavel Machek <pavel@denx.de>
> > 
> > diff --git a/fs/fs.c b/fs/fs.c
> > index 79d432d..ea15c5f 100644
> > --- a/fs/fs.c
> > +++ b/fs/fs.c
> > @@ -276,6 +276,7 @@ int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
> >  	unsigned long pos;
> >  	int len_read;
> >  	unsigned long time;
> > +	char *ep;
> >  
> >  	if (argc < 2)
> >  		return CMD_RET_USAGE;
> > @@ -286,7 +287,9 @@ int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
> >  		return 1;
> >  
> >  	if (argc >= 4) {
> > -		addr = simple_strtoul(argv[3], NULL, 16);
> > +		addr = simple_strtoul(argv[3], &ep, 16);
> > +		if (ep == argv[3] || *ep != '\0')
> > +			return CMD_RET_USAGE;
> 
> What happens in case of filenames that look like numbers, say "0"?
> It may be silly to use such, but it should not be impossible to do
> so.

It should be ok.

This catches error when someone passes non-number as a load
address. As long as you pass load address, nothing changes.

IOW  

fatload mmc 0:1 0 filename
# always worked/keeps working
fatload mmc 0:1 filename
# before my patch, this would try to parse filename as number, fail,
# and use default load address adn default filename. Not what user
# wanted.
fatload mmc 0:1 0 0
# Load filename "0" at address 0. Worked before, will work now.

Hope this helps,
									Pavel
Tom Rini July 22, 2014, 7:22 p.m. UTC | #6
On Wed, Jul 09, 2014 at 10:42:57PM +0200, Pavel Machek wrote:

> If filename is passed instead of address to ext2load or fatload,
> u-boot silently accepts that, and uses 0 for load address and default
> filename from environment. That is confusing, display help instead.
>     
> Signed-off-by: Pavel Machek <pavel@denx.de>

Applied to u-boot/master, thanks!
diff mbox

Patch

diff --git a/fs/fs.c b/fs/fs.c
index 79d432d..ea15c5f 100644
--- a/fs/fs.c
+++ b/fs/fs.c
@@ -276,6 +276,7 @@  int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
 	unsigned long pos;
 	int len_read;
 	unsigned long time;
+	char *ep;
 
 	if (argc < 2)
 		return CMD_RET_USAGE;
@@ -286,7 +287,9 @@  int do_load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[],
 		return 1;
 
 	if (argc >= 4) {
-		addr = simple_strtoul(argv[3], NULL, 16);
+		addr = simple_strtoul(argv[3], &ep, 16);
+		if (ep == argv[3] || *ep != '\0')
+			return CMD_RET_USAGE;
 	} else {
 		addr_str = getenv("loadaddr");
 		if (addr_str != NULL)