On Thu, Apr 04, 2019 at 04:30:38PM -0500, Eric Sandeen wrote:
> On 3/20/19 2:35 PM, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> >
> > Fix a number of complaints about feeding sizeof(dest) directly to
> > strncpy. We do this by declaring the char arrays to be one larger
> > than necessary and subtracting one, to ensure that we never overfill
> > the buffer.
> >
> > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> > ---
> > mkfs/xfs_mkfs.c | 13 +++++++++++--
> > quota/edit.c | 9 ++++++---
> > 2 files changed, 17 insertions(+), 5 deletions(-)
> >
> > diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> > index 9e1c6ec5..e87c692c 100644
> > --- a/mkfs/xfs_mkfs.c
> > +++ b/mkfs/xfs_mkfs.c
> > @@ -3251,8 +3251,17 @@ finish_superblock_setup(
> > struct xfs_mount *mp,
> > struct xfs_sb *sbp)
> > {
> > - if (cfg->label)
> > - strncpy(sbp->sb_fname, cfg->label, sizeof(sbp->sb_fname));
> > + if (cfg->label) {
> > + size_t label_len;
> > +
> > + /*
> > + * Labels are null terminated unless the string fits exactly
> > + * in the label field, so assume sb_fname is zeroed and then
> > + * do a memcpy because the destination isn't a normal C string.
> > + */
> > + label_len = min(sizeof(sbp->sb_fname), strlen(cfg->label));
>
> Heh, comparing this to the patch I wrote in <checks notes> December and never
> sent *cough*, it seems like it might be wise to memset the whole thing to 0,
> to be sure we're not leaking bytes here and there in a short label?
>
> + memset(sbp->sb_fname, 0, sizeof(sbp->sb_fname));
>
> Although, I suppose the prior strncpy had the same issue... *shrug*
I think the initialization is fine because...
finish_superblock_setup sets fields in *sbp. It gets *sbp from the top
of main(), which does:
struct xfs_mount mbuf = {};
struct xfs_mount *mp = &mbuf;
struct xfs_sb *sbp = &mp->m_sb;
So assuming the empty mbuf initializer causes the whole stack variable
to be initialized to zero, we don't need to do a memset here.
>
> > + memcpy(sbp->sb_fname, cfg->label, label_len);
> > + }
> >
> > sbp->sb_dblocks = cfg->dblocks;
> > sbp->sb_rblocks = cfg->rtblocks;
> > diff --git a/quota/edit.c b/quota/edit.c
> > index b10a5b34..f9938b8a 100644
> > --- a/quota/edit.c
> > +++ b/quota/edit.c
> > @@ -368,8 +368,7 @@ restore_file(
> > uint type)
> > {
> > char buffer[512];
> > - char devbuffer[512];
> > - char *dev = NULL;
> > + char dev[512];
> > uint mask;
> > int cnt;
> > uint32_t id;
> > @@ -377,7 +376,11 @@ restore_file(
> >
> > while (fgets(buffer, sizeof(buffer), fp) != NULL) {
> > if (strncmp("fs = ", buffer, 5) == 0) {
> > - dev = strncpy(devbuffer, buffer+5, sizeof(devbuffer));
> > + /*
> > + * Copy the device name to dev, strip off the trailing
> > + * newline, and move on to the next line.
> > + */
> > + strncpy(dev, buffer + 5, sizeof(dev) - 1);
> > dev[strlen(dev) - 1] = '\0';
> > continue;
> > }
>
> Ok.
>
> I had something in io/label.c: label_f() as well. Do you not get complaints
> about strncpy(label, optarg, sizeof(label)); ?
I do, that's why I changed it...? io/label.c got changed to shut up the
warning about the sizeof(firstarg) as the third parameter.
>
> I essentially did the same thing as for mkfs, though with my gratuitous
> memset:
>
> break;
> case 's':
> - strncpy(label, optarg, sizeof(label));
> + len = min(strlen(optarg), sizeof(label));
> + /* Kernel doesn't care about null termination */
> + memset(label, 0, sizeof(label));
> + memcpy(label, optarg, len);
> break;
> default:
>
(FWIW I'm not sure I remember what we were talking about here.)
Anyway, maybe I'll just resend the remaining patches as a rollup since
this series is already huge.
--D
