Author: nanxiao

Use pdb to help understand python program

As I have mentioned in Why do I need a debugger?:

(3) Debugger is a good tool to help you understand code.

So when I come across difficulty to understand vfscount.py code in bcc project, I know it is time to resort to pdb, python‘s debugger, to help me.

The thing which confuses me is here:

counts = b.get_table("counts")
for k, v in sorted(counts.items(), key=lambda counts: counts[1].value):
    print("%-16x %-26s %8d" % (k.ip, b.ksym(k.ip), v.value))

From previous code:

BPF_HASH(counts, struct key_t, u64, 256);

It seems the v‘s type is u64, and I can’t figure out why use v.value to fetch its data here.

pdb‘s manual is very succinct and its command is so similar with gdb‘s, so it is no learning curve for me. Just launch a debugging session and set breakpoint at “counts = b.get_table("counts")” line:

# python -m pdb vfscount.py
> /root/Project/bcc/tools/vfscount.py(14)<module>()
-> from __future__ import print_function
(Pdb) b vfscount.py:49

Start the program and press Ctrl-C after seconds; the breakpoint will be hit:

(Pdb) r
Tracing... Ctrl-C to end.
^C
ADDR             FUNC                          COUNT
> /root/Project/bcc/tools/vfscount.py(49)<module>()
-> counts = b.get_table("counts")

Step into get_table method, and single-step every line. Before leaving method, check the type of keytype and leaftype:

-> counts = b.get_table("counts")
(Pdb) s
--Call--
> /usr/lib/python3.6/site-packages/bcc/__init__.py(416)get_table()
-> def get_table(self, name, keytype=None, leaftype=None, reducer=None):
(Pdb) n
> /usr/lib/python3.6/site-packages/bcc/__init__.py(417)get_table()
-> map_id = lib.bpf_table_id(self.module, name.encode("ascii"))
......
(Pdb) p leaf_desc
b'"unsigned long long"'
(Pdb) n
> /usr/lib/python3.6/site-packages/bcc/__init__.py(430)get_table()
-> leaftype = BPF._decode_table_type(json.loads(leaf_desc.decode()))
(Pdb)
> /usr/lib/python3.6/site-packages/bcc/__init__.py(431)get_table()
-> return Table(self, map_id, map_fd, keytype, leaftype, reducer=reducer)
(Pdb) p leaftype
<class 'ctypes.c_ulong'>
(Pdb) p keytype
<class 'bcc.key_t'>

Yeah! The magic is here: leaftype‘s type is not pure u64, but ctypes.c_ulong. According to document:

>>> print(i.value)
42

We should use v.value to get its internal data.

Happy pdbing! Happy python debugging!

The timezone issue of installing OpenBSD in VirtualBox

I tried to use OpenBSD in VirtualBox. During installation, it prompted me:

Time appears wrong. Set to ‘Sat Aug 19 11:56:42 +08 2017’? [yes]

Since my server is UTC+8 timezone, I select “yes”. But after rebooting, the date command showed wrong info:

#date
Sat Aug 19 20:01:00 +8 2017
#date
Sat Aug 19 12:01:05 UTC 2017

Actually, my current host time should be 12:01:05 UTC+8. The VirtualBox seemed consider the host time as UTC time, and added another 8 hours. After discussing in the mailing list, the correct answer was found. I should tick “Hardware clock in UTC time” in setting:

Capture

According the manual:

If checked, VirtualBox will report the system time in UTC format to the guest instead of local (host) time. This affects how the virtual real-time clock (RTC) operates and may be useful for Unix-like guest operating systems, which typically expect the hardware clock to be set to UTC.

OpenBSD should consider time reported by VirtualBox as UTC time. So if this option is not checked, the VirtualBox will report local time to OpenBSD, OpenBSD misunderstands it as UTC time and add additional 8 hour to local time. This can explain what I have seen.

Reference:
Set date during OpenBSD installation.

 

Include “stdio.h” before Readline library header files

I install Readline library and write a simple program to play with it:

#include <readline/readline.h>
#include <readline/history.h>

int main(void)
{
    char *line_read = readline (">>  ");
    if (line_read && *line_read)
    {
        add_history (line_read);
    }
    ......
}

Build it will generate following errors:

In file included from /usr/include/readline/readline.h:35:0,
                 from readline.c:3:
/usr/include/readline/rltypedefs.h:71:28: error: unknown type name 'FILE'
 typedef int rl_getc_func_t PARAMS((FILE *));
                            ^
/usr/include/readline/readline.h:429:20: error: unknown type name 'FILE'
 extern int rl_getc PARAMS((FILE *));
                    ^
In file included from readline.c:3:0:
/usr/include/readline/readline.h:558:8: error: unknown type name 'FILE'
 extern FILE *rl_instream;
        ^~~~
/usr/include/readline/readline.h:559:8: error: unknown type name 'FILE'
 extern FILE *rl_outstream;
        ^~~~
/usr/include/readline/readline.h:588:8: error: unknown type name 'rl_getc_func_t'
 extern rl_getc_func_t *rl_getc_function;
        ^~~~~~~~~~~~~~
/usr/include/readline/readline.h:917:3: error: unknown type name 'FILE'
   FILE *inf;
   ^~~~
/usr/include/readline/readline.h:918:3: error: unknown type name 'FILE'
   FILE *outf;

The solution is including <stdio.h> before Readline library’s header files:

#include <stdio.h>
#include <readline/readline.h>
#include <readline/history.h>
......

Use Ctxt::isCorrect() to check Ctxt valid in using HElib

During using HElib, if the calculation made Ctxt‘s noise too big, it will cause Ctxt can’t be decrypted successfully, and the decrypted result is messy like this:

333317397173303 1016371194582039 217550837977960
737191953777559 1103851234911944 454685807668230
625213263895453 743713807221034 1036409711005678
447878070619549 603715795412661 948856958008786
......

You can leverage Ctxt::isCorrect() method to check whether Ctxt is still valid or not:

std::cout << "Ctxt is valid: " << c.isCorrect() << '\n';

Reference:
Maybe the bug in EncryptedArray’s shift function.

The subtleties of writing DTrace scripts

(1) Assume you write your DTrace script on Windows and run it on Unix. Because Windows uses \r\n as EOL while Unix uses \n, if you execute it directly, following complain will occur:

: No such file or directory

The solution is to apply dos2unix to convert file format before playing it.

(2) The shebang line of script only accepts 1 argument, and -s option must be the last. For example:

#!/usr/sbin/dtrace -sq

Will generate following error:

dtrace: failed to open q: No such file or directory

While “#!/usr/sbin/dtrace -qs” will be OK. To make additional options take effect, please use #pragma directive:

#pragma D option ...

Reference:
Does DTrace script only support “#pragma D option ..”, not “-x option”?.