[U-Boot] [PATCH v3 1/6] fat: write: fix broken write to fragmented files

[Marek Szyprowski]

Hi,

On 27.11.2019 03:26, AKASHI Takahiro wrote:
> Thank you for the heads-up.
>
> On Tue, Nov 26, 2019 at 11:57:29AM -0500, Tom Rini wrote:
>> On Tue, Nov 26, 2019 at 09:15:07AM +0100, Marek Szyprowski wrote:
>>
>>> The code for handing file overwrite incorrectly assumed that the file on
>>> disk is always contiguous. This resulted in corrupting disk structure
>>> every time when write to existing fragmented file happened. Fix this
>>> by adding proper check for cluster discontinuity and adjust chunk size
>>> on each partial write.
>>>
>>> Signed-off-by: Marek Szyprowski <m.szyprowski at samsung.com>
>>> Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov at toradex.com>
>>> Reviewed-by: Lukasz Majewski <lukma at denx.de>
>>> ---
>>>   fs/fat/fat_write.c | 6 +++---
>>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/fs/fat/fat_write.c b/fs/fat/fat_write.c
>>> index 729cf39630..6cfa5b4565 100644
>>> --- a/fs/fat/fat_write.c
>>> +++ b/fs/fat/fat_write.c
>>> @@ -794,6 +794,8 @@ set_contents(fsdata *mydata, dir_entry *dentptr, loff_t pos, __u8 *buffer,
>>>   
>>>   			newclust = get_fatent(mydata, endclust);
>>>   
>>> +			if ((newclust - 1) != endclust)
> "newclust != (endclust + 1)" would be more intuitive?
> But it's just my preference.

Indeed.

>>> +				break;
>>>   			if (IS_LAST_CLUST(newclust, mydata->fatsize))
>>>   				break;
>>>   			if (CHECK_CLUST(newclust, mydata->fatsize)) {
>>> @@ -811,7 +813,7 @@ set_contents(fsdata *mydata, dir_entry *dentptr, loff_t pos, __u8 *buffer,
>>>   			offset = 0;
>>>   		else
>>>   			offset = pos - cur_pos;
>>> -		wsize = min(cur_pos + actsize, filesize) - pos;
>>> +		wsize = min_t(unsigned long long, actsize, filesize - cur_pos);
> This hunk is not directly related to the issue, is it?

It is partially related. I remember that it was not calculated correctly 
for the fragmented files and then discovered that there was one more 
case in which the current formula failed.

>>>   		if (get_set_cluster(mydata, curclust, offset,
>>>   				    buffer, wsize, &actsize)) {
>>>   			printf("Error get-and-setting cluster\n");
>>> @@ -824,8 +826,6 @@ set_contents(fsdata *mydata, dir_entry *dentptr, loff_t pos, __u8 *buffer,
>>>   		if (filesize <= cur_pos)
>>>   			break;
>>>   
>>> -		/* CHECK: newclust = get_fatent(mydata, endclust); */
>>> -
>>>   		if (IS_LAST_CLUST(newclust, mydata->fatsize))
>>>   			/* no more clusters */
>>>   			break;
>> Adding in Heinrich and Akashi-san for more review on this, thanks!
> Otherwise, it looks good.
> Reviewed-by: AKASHI Takahiro <takahiro.akashi at linaro.org>

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland