linux-kernel - Re: [PATCHv2 6/9] staging: vme_user: return -EFAULT on __copy_*

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 25 Jun 2015 20:05:08 +0300
From:	Dmitry Kalinkin <dmitry.kalinkin@...il.com>
To:	Sudip Mukherjee <sudipm.mukherjee@...il.com>
Cc:	linux-kernel@...r.kernel.org, devel@...verdev.osuosl.org,
	Martyn Welch <martyn.welch@...com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Manohar Vanga <manohar.vanga@...il.com>
Subject: Re: [PATCHv2 6/9] staging: vme_user: return -EFAULT on __copy_*_user errors


> On 25 Jun 2015, at 15:05, Dmitry Kalinkin <dmitry.kalinkin@...il.com> wrote:
> 
> 
>> On 25 Jun 2015, at 14:27, Sudip Mukherjee <sudipm.mukherjee@...il.com> wrote:
>> 
>> On Tue, Jun 23, 2015 at 07:03:36PM +0300, Dmitry Kalinkin wrote:
>>> Signed-off-by: Dmitry Kalinkin <dmitry.kalinkin@...il.com>
>>> ---
>>> drivers/staging/vme/devices/vme_user.c | 47 ++++++++--------------------------
>>> 1 file changed, 11 insertions(+), 36 deletions(-)
>>> 
>> <snip>
>>> @@ -178,38 +167,24 @@ static ssize_t buffer_to_user(unsigned int minor, char __user *buf,
>>> 			      size_t count, loff_t *ppos)
>>> {
>>> 	void *image_ptr;
>>> -	ssize_t retval;
>>> 
>>> 	image_ptr = image[minor].kern_buf + *ppos;
>>> +	if (__copy_to_user(buf, image_ptr, (unsigned long)count))
>>> +		return -EFAULT;
>>> 
>>> -	retval = __copy_to_user(buf, image_ptr, (unsigned long)count);
>>> -	if (retval != 0) {
>>> -		retval = (count - retval);
>>> -		pr_warn("Partial copy to userspace\n");
>>> -	} else
>>> -		retval = count;
>>> -
>>> -	/* Return number of bytes successfully read */
>>> -	return retval;
>>> +	return count;
>> will it not affect the userspace code?
>> previously number of bytes successfully read was returned, now incase of
>> partial read -EFAULT is being returned.
> Exactly.
> 
> Practically there is an access_ok() call in vfs_read() and vfs_write() that
> will catch this first.  I don’t know exactly what is the condition for
> __copy_to_user to fail, but it is probably some rare arch-specific thing (and
> we only care for x86/powerpc here). But when it happens it better be returning
> proper error codes. This is why I think this is not a “we broke userspace”
> situation.

It seems like what I wrote above is not correct. access_ok does only coarse
checks and __copy_to_user does fail. Anyway, only a rare userspace application
would depend on “succesfull” read that was interrupted by a segfault. Also, if
__copy_to_user fails completely, the original code would return zero, which in
POSIX should mean something like “everything is good, try again later” and
this may cause infinite loops (e.g. python).--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/