Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 26 Feb 2010 11:21:05 -0500
From:      John Baldwin <jhb@freebsd.org>
To:        freebsd-hackers@freebsd.org
Cc:        freebsd-emulation@freebsd.org, Tim Kientzle <kientzle@freebsd.org>, Juergen Lock <nox@jelal.kn-bremen.de>
Subject:   Re: 32 bit Linux lseek missing overflow check (was: Re: Linuxolator patches: stat and lseek SEEK_END for disk devices)
Message-ID:  <201002261121.05068.jhb@freebsd.org>
In-Reply-To: <20100225202850.GA79505@triton8.kn-bremen.de>
References:  <20100223215010.GA67619@triton8.kn-bremen.de> <20100225202850.GA79505@triton8.kn-bremen.de>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday 25 February 2010 3:28:50 pm Juergen Lock wrote:
> On Tue, Feb 23, 2010 at 10:50:10PM +0100, Juergen Lock wrote:
> > Hi!
> > 
> >  Before this gets buried on -hackers in another thead... :)
> > 
> >  I now have disks appear as block devices for Linux processes (there
> > already was commented out code for that in linux_stats.c, I hope my
> > version is now `correct enough' to be usable [1]), and I made a simple
> > patch to make lseek SEEK_END (L_XTND in the source) dtrt on disk
> > devices too by simply invoking the DIOCGMEDIASIZE ioctl there; [2]
> > both of these things are what (some) Linux processes expect.
> > 
> >  Patches are here: (made on stable/8, if they don't apply on head
> > I'll have to make extra versions for that...)
> > 	http://people.freebsd.org/~nox/linuxdisk-blk.patch [1]
> > 	http://people.freebsd.org/~nox/lseek-seek_end.patch [2]
> > 
> >  And yes, with these patches the Linux bsdtar mentioned on -hackers
> > in the `"tar tfv /dev/cd0" speedup patch' thread now also runs fast
> > on FreeBSD. :)
> 
> I now added an vn_isdisk() check to the second patch after comments from
> julian, and I made a new patch that adds an overflow check to the 32 bit
> linux lseek: (also at
> 	http://people.freebsd.org/~nox/linux-lseek-overflow.patch
> )

Hmm, when I asked Bruce, he actually said it was a feature that you didn't use 
vn_isdisk().  He also suggested that the proper way to fix lseek on devices is 
to fix stat in devfs to return a non-zero size.  I have a possible fix to do 
that but haven't tested it yet:

Index: devfs_vnops.c
===================================================================
--- devfs_vnops.c	(revision 204207)
+++ devfs_vnops.c	(working copy)
@@ -44,6 +44,7 @@
 #include <sys/systm.h>
 #include <sys/conf.h>
 #include <sys/dirent.h>
+#include <sys/disk.h>
 #include <sys/fcntl.h>
 #include <sys/file.h>
 #include <sys/filedesc.h>
@@ -564,7 +565,11 @@
 	struct vattr *vap = ap->a_vap;
 	int error = 0;
 	struct devfs_dirent *de;
+	struct cdevsw *dsw;
+	struct thread *td;
+	struct file *fpop;
 	struct cdev *dev;
+	off_t size;
 
 	de = vp->v_data;
 	KASSERT(de != NULL, ("Null dirent in devfs_getattr vp=%p", vp));
@@ -612,6 +617,18 @@
 		vap->va_ctime = dev->si_ctime;
 
 		vap->va_rdev = cdev2priv(dev)->cdp_inode;
+
+		dsw = dev_refthread(dev);
+		if (dsw != NULL) {
+			td = curthread;
+			fpop = td->td_fpop;
+			td->td_fpop = NULL;
+			if (dsw->d_ioctl(dev, DIOCGMEDIASIZE, (caddr_t)&size,
+			    FREAD, td) == 0)
+				vap->va_size = size;
+			td->td_fpop = fpop;
+			dev_relthread(dev);
+		}
 	}
 	vap->va_gen = 0;
 	vap->va_flags = 0;


-- 
John Baldwin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201002261121.05068.jhb>