From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH 14/27] drivers/scsi: Use memdup_user Date: Sat, 22 May 2010 09:24:23 -0500 Message-ID: <1274538263.4410.38.camel@mulgrave.site> References: Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Julia Lawall Cc: Doug Gilbert , linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, kernel-janitors@vger.kernel.org List-Id: linux-scsi@vger.kernel.org On Sat, 2010-05-22 at 10:22 +0200, Julia Lawall wrote: > From: Julia Lawall > > Use memdup_user when user data is immediately copied into the > allocated region. > > The semantic patch that makes this change is as follows: > (http://coccinelle.lip6.fr/) > > // > @@ > expression from,to,size,flag; > position p; > identifier l1,l2; > @@ > > - to = \(kmalloc@p\|kzalloc@p\)(size,flag); > + to = memdup_user(from,size); > if ( > - to==NULL > + IS_ERR(to) > || ...) { > <+... when != goto l1; > - -ENOMEM > + PTR_ERR(to) > ...+> > } > - if (copy_from_user(to, from, size) != 0) { > - <+... when != goto l2; > - -EFAULT > - ...+> > - } > // > > Signed-off-by: Julia Lawall > > --- > drivers/scsi/sg.c | 11 +++-------- > 1 file changed, 3 insertions(+), 8 deletions(-) > > diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c > index ef752b2..277ace6 100644 > --- a/drivers/scsi/sg.c > +++ b/drivers/scsi/sg.c > @@ -1676,14 +1676,9 @@ static int sg_start_req(Sg_request *srp, unsigned char *cmd) > int len, size = sizeof(struct sg_iovec) * iov_count; > struct iovec *iov; > > - iov = kmalloc(size, GFP_ATOMIC); > - if (!iov) > - return -ENOMEM; > - > - if (copy_from_user(iov, hp->dxferp, size)) { > - kfree(iov); > - return -EFAULT; > - } > + iov = memdup_user(hp->dxferp, size); > + if (IS_ERR(iov)) > + return PTR_ERR(iov); This type of transformation has really no value at all. The code you're proposing to replace is already correct. I'm fairly ambivalent on patterned APIs anyway but I accept they're useful way to prevent new code getting it wrong. However, it's completely bogus to force replacement of correctly functioning code throughout the kernel (unless you're going to argue that everyone who tries to copy from user into a kmalloc space does a cut and paste from sg?) Of infinitely greater service would be finding any places where the pattern is being incorrectly used. So, if Doug wants to take this as a prettify of sg, I'm happy, but if not, I don't really want to see this again. Thanks, James