Matt Helsley wrote: > Unlike the length passed into ckpt_write_obj_type, the maximum length passed > to ckpt_read_buf_type must include the length of the struct ckpt_hdr. IMHO, the right way to fix this is to change ckpt_read_obj_type(). This will preserve symmetry between checkpoint and restart, and also fix a similar problem in kernel/groups.c (MAX_GROUPINFO_SIZE). No need to resend - I'll fix already. Oren. > > Signed-off-by: Matt Helsley <matthltc@xxxxxxxxxx> > --- > checkpoint/files.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/checkpoint/files.c b/checkpoint/files.c > index 0564666..562c338 100644 > --- a/checkpoint/files.c > +++ b/checkpoint/files.c > @@ -449,7 +449,7 @@ struct file *restore_open_fname(struct ckpt_ctx *ctx, int flags) > if (flags & (O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC)) > return ERR_PTR(-EINVAL); > > - h = ckpt_read_buf_type(ctx, PATH_MAX, CKPT_HDR_FILE_NAME); > + h = ckpt_read_buf_type(ctx, PATH_MAX + sizeof(*h), CKPT_HDR_FILE_NAME); > if (IS_ERR(h)) > return (struct file *) h; > len = h->len - sizeof(*h); _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers