linux-kernel - Re: [Qemu-devel] [PATCH 6/7] qemu: Implement virtio-pstore device

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] [day] [month] [year] [list]

Message-ID: <20160801155255.GB24526@danjae.aot.lge.com>
Date:	Tue, 2 Aug 2016 00:52:55 +0900
From:	Namhyung Kim <namhyung@...nel.org>
To:	"Daniel P. Berrange" <berrange@...hat.com>
Cc:	kvm@...r.kernel.org, qemu-devel@...gnu.org,
	virtualization@...ts.linux-foundation.org,
	Tony Luck <tony.luck@...el.com>,
	Radim Kr??m???? <rkrcmar@...hat.com>,
	Kees Cook <keescook@...omium.org>,
	"Michael S. Tsirkin" <mst@...hat.com>,
	Anton Vorontsov <anton@...msg.org>,
	LKML <linux-kernel@...r.kernel.org>,
	Steven Rostedt <rostedt@...dmis.org>,
	Minchan Kim <minchan@...nel.org>,
	Anthony Liguori <aliguori@...zon.com>,
	Colin Cross <ccross@...roid.com>,
	Paolo Bonzini <pbonzini@...hat.com>,
	Ingo Molnar <mingo@...nel.org>
Subject: Re: [Qemu-devel] [PATCH 6/7] qemu: Implement virtio-pstore device

On Mon, Aug 01, 2016 at 10:24:39AM +0100, Daniel P. Berrange wrote:
> On Sat, Jul 30, 2016 at 05:57:02PM +0900, Namhyung Kim wrote:
> > On Thu, Jul 28, 2016 at 02:22:39PM +0100, Daniel P. Berrange wrote:
> > > > +static void virtio_pstore_from_filename(VirtIOPstore *s, char *name,
> > > > +                                        char *buf, size_t sz,
> > > > +                                        struct virtio_pstore_fileinfo *info)
> > > > +{
> > > > +    snprintf(buf, sz, "%s/%s", s->directory, name);
> > > > +
> > > > +    if (g_str_has_prefix(name, "dmesg-")) {
> > > > +        info->type = VIRTIO_PSTORE_TYPE_DMESG;
> > > > +        name += strlen("dmesg-");
> > > > +    } else if (g_str_has_prefix(name, "console-")) {
> > > > +        info->type = VIRTIO_PSTORE_TYPE_CONSOLE;
> > > > +        name += strlen("console-");
> > > > +    } else if (g_str_has_prefix(name, "unknown-")) {
> > > > +        info->type = VIRTIO_PSTORE_TYPE_UNKNOWN;
> > > > +        name += strlen("unknown-");
> > > > +    }
> 
> [snip]
> 
> > > > +    struct virtio_pstore_fileinfo info;
> > > > +    size_t offset = sizeof(*res) + sizeof(info);
> > > > +
> > > > +    if (s->dirp == NULL) {
> > > > +        return -1;
> > > > +    }
> > > > +
> > > > +    dent = readdir(s->dirp);
> > > > +    while (dent) {
> > > > +        if (dent->d_name[0] != '.') {
> > > > +            break;
> > > > +        }
> > > > +        dent = readdir(s->dirp);
> > > > +    }
> > > > +
> > > > +    if (dent == NULL) {
> > > > +        return 0;
> > > > +    }
> > > 
> > > So this seems to just be picking the first filename reported by
> > > readdir that isn't starting with '.'. Surely this can't the right
> > > logic when your corresponding do_write method can pick several
> > > different filenames, its potluck which do_read will give back.
> > 
> > Do you mean that it'd be better to check a list of known filenames and
> > fail if not?
> 
> No, I mean that you have several different VIRTIO_PSTORE_TYPE_nnn and
> use a different file for each constant. When reading this directory
> though you're not looking for the file corresponding to any given
> VIRTIO_PSTORE_TYPE_nnn - you're simply reading whichever file is found
> first. So you might have just read a TYPE_CONSOLE or have read a
> TYPE_DMESG - it surely doesn't make sense to randonly read either
> TYPE_CONSOLE or TYPE_DMESG based on whatever order readdir() lists
> the files.

In fact, each VIRTIO_PSTORE_TYPE_xxx can have more than one files and
I don't care about the ordering between them.  They will be fed to the
pstore filesystem on the guest.

But I also think it'd be better to scan files of known types only
rather than read out whatever readdir() gives.

Thanks,
Namhyung