linux-kernel - Re: [PATCH v2] scripts/gdb: fix lx-dmesg when CONFIG_PRINTK

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <VI1PR04MB70239DA9EED5F689645071E9EE970@VI1PR04MB7023.eurprd04.prod.outlook.com>
Date:   Fri, 11 Oct 2019 12:47:26 +0000
From:   Leonard Crestez <leonard.crestez@....com>
To:     Joel Colledge <joel.colledge@...bit.com>,
        Jan Kiszka <jan.kiszka@...mens.com>
CC:     Kieran Bingham <kbingham@...nel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] scripts/gdb: fix lx-dmesg when CONFIG_PRINTK_CALLER is
 set

On 11.10.2019 15:25, Joel Colledge wrote:
> When CONFIG_PRINTK_CALLER is set, struct printk_log contains an
> additional member caller_id. This affects the offset of the log text.
> Account for this by using the type information from gdb to determine all
> the offsets instead of using hardcoded values.
> 
> This fixes following error:
> 
>    (gdb) lx-dmesg
>    Python Exception <class 'ValueError'> embedded null character:
>    Error occurred in Python command: embedded null character
> 
> Signed-off-by: Joel Colledge <joel.colledge@...bit.com>
> ---
> Changes in v2:
> - use type information from gdb instead of hardcoded offsets
> 
> Thanks for the idea about using the struct layout info from gdb, Leonard. I can't see any reason we shouldn't use that here, since most of the other commands use it. LxDmesg has used hardcoded offsets since scripts/gdb was introduced, so I assume it just ended up like that during the initial development of the tool. Here is a version of the fix using offsets from gdb.

This struct printk_log is quite small, I wonder if it's possible to do a 
single read for each log entry? This might make lx-dmesg faster because 
of fewer roundtrips to gdbserver and jtag (or whatever backend you're 
using).

> 
>   scripts/gdb/linux/dmesg.py | 16 ++++++++++++----
>   1 file changed, 12 insertions(+), 4 deletions(-)
> 
> diff --git a/scripts/gdb/linux/dmesg.py b/scripts/gdb/linux/dmesg.py
> index 6d2e09a2ad2f..8f5d899029b7 100644
> --- a/scripts/gdb/linux/dmesg.py
> +++ b/scripts/gdb/linux/dmesg.py
> @@ -16,6 +16,8 @@ import sys
>   
>   from linux import utils
>   
> +printk_log_type = utils.CachedType("struct printk_log")
> +
>   
>   class LxDmesg(gdb.Command):
>       """Print Linux kernel log buffer."""
> @@ -42,9 +44,14 @@ class LxDmesg(gdb.Command):
>               b = utils.read_memoryview(inf, log_buf_addr, log_next_idx)
>               log_buf = a.tobytes() + b.tobytes()
>   
> +        length_offset = printk_log_type.get_type()['len'].bitpos // 8
> +        text_len_offset = printk_log_type.get_type()['text_len'].bitpos // 8
> +        time_stamp_offset = printk_log_type.get_type()['ts_nsec'].bitpos // 8
> +        text_offset = printk_log_type.get_type().sizeof
> +
>           pos = 0
>           while pos < log_buf.__len__():
> -            length = utils.read_u16(log_buf[pos + 8:pos + 10])
> +            length = utils.read_u16(log_buf[pos + length_offset:pos + length_offset + 2])
>               if length == 0:
>                   if log_buf_2nd_half == -1:
>                       gdb.write("Corrupted log buffer!\n")
> @@ -52,10 +59,11 @@ class LxDmesg(gdb.Command):
>                   pos = log_buf_2nd_half
>                   continue
>   
> -            text_len = utils.read_u16(log_buf[pos + 10:pos + 12])
> -            text = log_buf[pos + 16:pos + 16 + text_len].decode(
> +            text_len = utils.read_u16(log_buf[pos + text_len_offset:pos + text_len_offset + 2])
> +            text = log_buf[pos + text_offset:pos + text_offset + text_len].decode(
>                   encoding='utf8', errors='replace')
> -            time_stamp = utils.read_u64(log_buf[pos:pos + 8])
> +            time_stamp = utils.read_u64(
> +                log_buf[pos + time_stamp_offset:pos + time_stamp_offset + 8])
>   
>               for line in text.splitlines():
>                   msg = u"[{time:12.6f}] {line}\n".format(
>