From: James Bottomley <James.Bottomley@HansenPartnership.com>
To: Alan Stern <stern@rowland.harvard.edu>
Cc: Boaz Harrosh <bharrosh@panasas.com>,
"Martin K. Petersen" <martin.petersen@oracle.com>,
Matthew Wilcox <willy@linux.intel.com>,
SCSI development list <linux-scsi@vger.kernel.org>
Subject: Re: Bugs in scsi_vpd_inquiry()
Date: Tue, 11 Aug 2009 10:59:09 -0500 [thread overview]
Message-ID: <1250006349.4301.46.camel@mulgrave.site> (raw)
In-Reply-To: <Pine.LNX.4.44L0.0908111136350.2562-100000@iolanthe.rowland.org>
On Tue, 2009-08-11 at 11:38 -0400, Alan Stern wrote:
> On Tue, 11 Aug 2009, James Bottomley wrote:
>
> > > > This is pointless and dangerous: Some architectures will invalidate
> > > > caches for DMA not flush them, so it might not do what you think it
> > > > does.
> > > >
> > >
> > > Then you can't do that "after check" either. It's a simple minefield.
> > > What do you suggest?
> >
> > Well, nothing really, like the original code. Byzantine checking is
> > never a good idea. You always assume that if a device told you it did
> > what you told it, it actually did. If we find devices that fail this
> > simple premise, then we get into blacklists and other forms of
> > nastiness.
>
> Okay, then how about this?
>
> Alan Stern
>
>
>
> Index: usb-2.6/drivers/scsi/scsi.c
> ===================================================================
> --- usb-2.6.orig/drivers/scsi/scsi.c
> +++ usb-2.6/drivers/scsi/scsi.c
> @@ -980,7 +980,8 @@ static int scsi_vpd_inquiry(struct scsi_
> u8 page, unsigned len)
> {
> int result;
> - unsigned char cmd[16];
> + int resid;
> + unsigned char cmd[6];
>
> cmd[0] = INQUIRY;
> cmd[1] = 1; /* EVPD */
> @@ -994,12 +995,12 @@ static int scsi_vpd_inquiry(struct scsi_
> * all the existing users tried this hard.
> */
> result = scsi_execute_req(sdev, cmd, DMA_FROM_DEVICE, buffer,
> - len + 4, NULL, 30 * HZ, 3, NULL);
> + len, NULL, 30 * HZ, 3, &resid);
> if (result)
> return result;
>
> - /* Sanity check that we got the page back that we asked for */
> - if (buffer[1] != page)
> + /* Sanity check that we at least got the header */
> + if (resid > len - 4)
> return -EIO;
>
> return 0;
> @@ -1027,7 +1028,7 @@ unsigned char *scsi_get_vpd_page(struct
> return NULL;
>
> /* Ask for all the pages supported by this device */
> - result = scsi_vpd_inquiry(sdev, buf, 0, 255);
> + result = scsi_vpd_inquiry(sdev, buf, 0, 255 + 4);
> if (result)
> goto fail;
>
> @@ -1042,7 +1043,7 @@ unsigned char *scsi_get_vpd_page(struct
> goto fail;
>
> found:
> - result = scsi_vpd_inquiry(sdev, buf, page, 255);
> + result = scsi_vpd_inquiry(sdev, buf, page, 255 + 4);
> if (result)
> goto fail;
>
> @@ -1056,7 +1057,7 @@ unsigned char *scsi_get_vpd_page(struct
>
> kfree(buf);
> buf = kmalloc(len + 4, GFP_KERNEL);
> - result = scsi_vpd_inquiry(sdev, buf, page, len);
> + result = scsi_vpd_inquiry(sdev, buf, page, len + 4);
> if (result)
> goto fail;
Sort of, but it's not really doing it properly. Lets do it like this.
This should also fix the > 255 length problem older devices might have.
James
---
diff --git a/drivers/scsi/scsi.c b/drivers/scsi/scsi.c
index 2de5f3a..b6e0307 100644
--- a/drivers/scsi/scsi.c
+++ b/drivers/scsi/scsi.c
@@ -994,7 +994,7 @@ static int scsi_vpd_inquiry(struct scsi_device *sdev, unsigned char *buffer,
* all the existing users tried this hard.
*/
result = scsi_execute_req(sdev, cmd, DMA_FROM_DEVICE, buffer,
- len + 4, NULL, 30 * HZ, 3, NULL);
+ len, NULL, 30 * HZ, 3, NULL);
if (result)
return result;
@@ -1021,13 +1021,14 @@ unsigned char *scsi_get_vpd_page(struct scsi_device *sdev, u8 page)
{
int i, result;
unsigned int len;
- unsigned char *buf = kmalloc(259, GFP_KERNEL);
+ const unsigned int init_vpd_len = 255;
+ unsigned char *buf = kmalloc(init_vpd_len, GFP_KERNEL);
if (!buf)
return NULL;
/* Ask for all the pages supported by this device */
- result = scsi_vpd_inquiry(sdev, buf, 0, 255);
+ result = scsi_vpd_inquiry(sdev, buf, 0, init_vpd_len);
if (result)
goto fail;
@@ -1050,12 +1051,12 @@ unsigned char *scsi_get_vpd_page(struct scsi_device *sdev, u8 page)
* Some pages are longer than 255 bytes. The actual length of
* the page is returned in the header.
*/
- len = (buf[2] << 8) | buf[3];
- if (len <= 255)
+ len = ((buf[2] << 8) | buf[3]) + 4;
+ if (len <= init_vpd_len)
return buf;
kfree(buf);
- buf = kmalloc(len + 4, GFP_KERNEL);
+ buf = kmalloc(len, GFP_KERNEL);
result = scsi_vpd_inquiry(sdev, buf, page, len);
if (result)
goto fail;
next prev parent reply other threads:[~2009-08-11 15:59 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-08-10 14:41 Bugs in scsi_vpd_inquiry() Alan Stern
2009-08-10 14:58 ` Matthew Wilcox
2009-08-10 15:32 ` Alan Stern
2009-08-10 17:08 ` Martin K. Petersen
2009-08-10 20:13 ` Alan Stern
2009-08-10 20:49 ` Martin K. Petersen
2009-08-10 21:14 ` Alan Stern
2009-08-10 22:47 ` Martin K. Petersen
2009-08-11 14:35 ` Alan Stern
2009-08-10 21:53 ` Douglas Gilbert
2009-08-10 22:52 ` Martin K. Petersen
2009-08-11 16:04 ` Matthew Wilcox
2009-08-11 7:07 ` Boaz Harrosh
2009-08-11 14:53 ` Alan Stern
2009-08-11 15:13 ` James Bottomley
2009-08-11 15:18 ` Boaz Harrosh
2009-08-11 15:27 ` James Bottomley
2009-08-11 15:38 ` Alan Stern
2009-08-11 15:57 ` Matthew Wilcox
2009-08-11 15:59 ` James Bottomley [this message]
2009-08-11 16:14 ` Alan Stern
2009-08-11 16:24 ` James Bottomley
2009-08-13 13:58 ` Boaz Harrosh
2009-08-13 14:15 ` James Bottomley
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1250006349.4301.46.camel@mulgrave.site \
--to=james.bottomley@hansenpartnership.com \
--cc=bharrosh@panasas.com \
--cc=linux-scsi@vger.kernel.org \
--cc=martin.petersen@oracle.com \
--cc=stern@rowland.harvard.edu \
--cc=willy@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox