| 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
| ||
|
Message-ID: <20191022170340.GA4641@hsiangkao-HP-ZHAN-66-Pro-G1>
Date: Wed, 23 Oct 2019 01:03:47 +0800
From: Gao Xiang <hsiangkao@....com>
To: Pratik Shinde <pratikshinde320@...il.com>
Cc: linux-erofs@...ts.ozlabs.org, Gao Xiang <gaoxiang25@...wei.com>,
yuchao0@...wei.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH-v2] erofs: code for verifying superblock checksum of an
erofs image.
On Tue, Oct 22, 2019 at 10:18:23PM +0530, Pratik Shinde wrote:
> Hi Gao,
>
> Understood your concern.
>
> Can we do something like :
>
> 1) Allocate one buf of size EROFS_BLKSIZ
> 2) Read one page at a time into buf(memcpy) .call crc32c for it.
>
> In this way we won't be writing directly into page data and will not do
> large allocation.
> What do you think?
Yes, that is the right way, but the buf is only be used to calc
the 0th block (super_block), for the rest blocks, no need to do this.
buf = kmalloc(EROFS_BLKSIZ);
memcpy(buf, data);
dsb->checksum = 0;
crc = crc32c(0, buf);
kfree(buf);
for (i = 1; i < nblocks; ++i) {
page = get_meta_page(sb, i);
data = kmap_atomic(page);
crc = crc32c(crc, data);
kunmap_atomic(data);
unlock_page(page);
put_page(page);
}
...
Or some better approach, but it's not ok to modify page cache directly.
Thanks,
Gao Xiang
>
> --Pratik
>
> On Tue, 22 Oct, 2019, 10:04 PM Gao Xiang, <hsiangkao@....com> wrote:
>
> > Hi Pratik,
> >
> > Some comments as below...
> >
> > On Tue, Oct 22, 2019 at 09:09:56PM +0530, Pratik Shinde wrote:
> > > Patch for kernel side changes of checksum feature.Used kernel's
> > > crc32c library for calculating the checksum.
> > >
> > > Signed-off-by: Pratik Shinde <pratikshinde320@...il.com>
> > > ---
> > > fs/erofs/erofs_fs.h | 5 +++--
> > > fs/erofs/internal.h | 3 ++-
> > > fs/erofs/super.c | 33 +++++++++++++++++++++++++++++++++
> > > 3 files changed, 38 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
> > > index b1ee565..4d8097a 100644
> > > --- a/fs/erofs/erofs_fs.h
> > > +++ b/fs/erofs/erofs_fs.h
> > > @@ -17,6 +17,7 @@
> > > */
> > > #define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING 0x00000001
> > > #define EROFS_ALL_FEATURE_INCOMPAT
> > EROFS_FEATURE_INCOMPAT_LZ4_0PADDING
> > > +#define EROFS_FEATURE_COMPAT_SB_CHKSUM 0x00000001
> > >
> > > /* 128-byte erofs on-disk super block */
> > > struct erofs_super_block {
> > > @@ -37,8 +38,8 @@ struct erofs_super_block {
> > > __u8 uuid[16]; /* 128-bit uuid for volume */
> > > __u8 volume_name[16]; /* volume name */
> > > __le32 feature_incompat;
> > > -
> > > - __u8 reserved2[44];
> > > + __le32 chksum_blocks; /* number of blocks used for checksum */
> > > + __u8 reserved2[40];
> > > };
> > >
> > > /*
> > > diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
> > > index 544a453..cec27ca 100644
> > > --- a/fs/erofs/internal.h
> > > +++ b/fs/erofs/internal.h
> > > @@ -86,7 +86,7 @@ struct erofs_sb_info {
> > > u8 uuid[16]; /* 128-bit uuid for volume */
> > > u8 volume_name[16]; /* volume name */
> > > u32 feature_incompat;
> > > -
> > > + u32 feature_compat;
> > > unsigned int mount_opt;
> > > };
> > >
> > > @@ -426,6 +426,7 @@ static inline void z_erofs_exit_zip_subsystem(void)
> > {}
> > > #endif /* !CONFIG_EROFS_FS_ZIP */
> > >
> > > #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
> > > +#define EFSBADCRC EBADMSG /* Bad crc found */
> > >
> > > #endif /* __EROFS_INTERNAL_H */
> > >
> > > diff --git a/fs/erofs/super.c b/fs/erofs/super.c
> > > index 0e36949..9cda72d 100644
> > > --- a/fs/erofs/super.c
> > > +++ b/fs/erofs/super.c
> > > @@ -9,6 +9,7 @@
> > > #include <linux/statfs.h>
> > > #include <linux/parser.h>
> > > #include <linux/seq_file.h>
> > > +#include <linux/crc32c.h>
> > > #include "xattr.h"
> > >
> > > #define CREATE_TRACE_POINTS
> > > @@ -46,6 +47,31 @@ void _erofs_info(struct super_block *sb, const char
> > *function,
> > > va_end(args);
> > > }
> > >
> > > +static int erofs_validate_sb_chksum(struct erofs_super_block *dsb,
> > > + struct super_block *sb)
> > > +{
> > > + u32 disk_chksum, nblocks, crc = 0;
> > > + void *kaddr;
> > > + struct page *page;
> > > + int i;
> > > +
> > > + disk_chksum = le32_to_cpu(dsb->checksum);
> > > + nblocks = le32_to_cpu(dsb->chksum_blocks);
> >
> > We cannot write the page data directly since the page cache should be kept
> > in
> > sync with ondisk data (or for read-write fs, if it's claimed as uptodated,
> > and
> > it is modified later, you should mark it dirty, and do writeback then, but
> > that is not the erofs case.)
> >
> > > + dsb->checksum = 0;
> > > + for (i = 0; i < nblocks; i++) {
> > > + page = erofs_get_meta_page(sb, i);
> > > + if (IS_ERR(page))
> > > + return PTR_ERR(page);
> > > + kaddr = kmap(page);
> >
> > Here kmap_atomic(page) is better. what I mean is kmap_atomic() in the
> > caller
> > erofs_read_superblock(), it should be replaced to kmap() instead.
> >
> > > + crc = crc32c(crc, kaddr, EROFS_BLKSIZ);
> > > + kunmap(page);
> > > + unlock_page(page);
> >
> > need
> > put_page(page);
> >
> >
> > I'm not sure whether I explained quite well, but this patch needs something
> > to do. I'm now working on demonstrating new XZ algorithm and releasing
> > erofs-utils v1.0.
> >
> > You can give more tries or I will help later. :-)
> >
> > Thanks,
> > Gao Xiang
> >
> >
> > > + }
> > > + if (crc != disk_chksum)
> > > + return -EFSBADCRC;
> > > + return 0;
> > > +}
> > > +
> > > static void erofs_inode_init_once(void *ptr)
> > > {
> > > struct erofs_inode *vi = ptr;
> > > @@ -121,6 +147,13 @@ static int erofs_read_superblock(struct super_block
> > *sb)
> > > goto out;
> > > }
> > >
> > > + if (dsb->feature_compat & EROFS_FEATURE_COMPAT_SB_CHKSUM) {
> > > + ret = erofs_validate_sb_chksum(dsb, sb);
> > > + if (ret < 0) {
> > > + erofs_err(sb, "super block checksum incorrect");
> > > + goto out;
> > > + }
> > > + }
> > > blkszbits = dsb->blkszbits;
> > > /* 9(512 bytes) + LOG_SECTORS_PER_BLOCK == LOG_BLOCK_SIZE */
> > > if (blkszbits != LOG_BLOCK_SIZE) {
> > > --
> > > 2.9.3
> > >
> >
Powered by blists - more mailing lists