[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c790aa59-7686-09e2-1066-92bec97704cd@redhat.com>
Date: Wed, 21 Aug 2019 10:17:09 -0500
From: Eric Sandeen <esandeen@...hat.com>
To: Lukas Czerner <lczerner@...hat.com>, linux-ext4@...r.kernel.org
Subject: Re: [PATCH 2/2] tune2fs: Warn if page size != blocksize when enabling
encrypt
On 8/21/19 8:18 AM, Lukas Czerner wrote:
> With encrypt feature enabled the file system block size must match
> system page size. Currently tune2fs will not complain at all when we try
> to enable encrypt on a file system that does not satisfy this
> requirement for the system. Add a warning for this case.
>
> Signed-off-by: Lukas Czerner <lczerner@...hat.com>
> ---
> misc/tune2fs.c | 38 ++++++++++++++++++++++++++++++++++++++
> 1 file changed, 38 insertions(+)
>
> diff --git a/misc/tune2fs.c b/misc/tune2fs.c
> index 7d2d38d7..26b1b1d0 100644
> --- a/misc/tune2fs.c
> +++ b/misc/tune2fs.c
> @@ -130,6 +130,8 @@ void do_findfs(int argc, char **argv);
> int journal_enable_debug = -1;
> #endif
>
> +static int sys_page_size = 4096;
> +
> static void usage(void)
> {
> fprintf(stderr,
> @@ -1407,6 +1409,29 @@ mmp_error:
> stderr);
> return 1;
> }
> +
> + /*
> + * When encrypt feature is enabled, the file system blocksize
> + * needs to match system page size otherwise the file system
> + * won't mount.
> + */
> + if (fs->blocksize != sys_page_size) {
> + if (!f_flag) {
> + com_err(program_name, 0,
> + _("Block size (%dB) does not match "
> + "system page size (%dB). File "
> + "system won't be usable on this "
> + "system"),
I wonder if this message should explicitly mention the encryption option, right
now it doesn't give a lot of context as to why this is being printed.
Perhaps "Encryption feature requested, but block size (%dB) does not match ...?"
-Eric
> + fs->blocksize, sys_page_size);
> + proceed_question(-1);
> + }
> + fprintf(stderr,_("Warning: Encrypt feature enabled, "
> + "but block size (%dB) does not match "
> + "system page size (%dB), forced to "
> + "cointinue\n"),
> + fs->blocksize, sys_page_size);
> + }
> +
> fs->super->s_encrypt_algos[0] =
> EXT4_ENCRYPTION_MODE_AES_256_XTS;
> fs->super->s_encrypt_algos[1] =
> @@ -2844,6 +2869,7 @@ int main(int argc, char **argv)
> int tune2fs_main(int argc, char **argv)
> #endif /* BUILD_AS_LIB */
> {
> + long sysval;
> errcode_t retval;
> ext2_filsys fs;
> struct ext2_super_block *sb;
> @@ -2879,6 +2905,18 @@ int tune2fs_main(int argc, char **argv)
> #endif
> io_ptr = unix_io_manager;
>
> + /* Determine the system page size if possible */
> +#ifdef HAVE_SYSCONF
> +#if (!defined(_SC_PAGESIZE) && defined(_SC_PAGE_SIZE))
> +#define _SC_PAGESIZE _SC_PAGE_SIZE
> +#endif
> +#ifdef _SC_PAGESIZE
> + sysval = sysconf(_SC_PAGESIZE);
> + if (sysval > 0)
> + sys_page_size = sysval;
> +#endif /* _SC_PAGESIZE */
> +#endif /* HAVE_SYSCONF */
> +
> retry_open:
> if ((open_flag & EXT2_FLAG_RW) == 0 || f_flag)
> open_flag |= EXT2_FLAG_SKIP_MMP;
>
Powered by blists - more mailing lists