resize hvm guest

$ dd if=/dev/zero of=new_image bs=1 count=new_size_of_ext3_in_bytes
$ dd if=old_image of=new_image conv=notrunc #copy the old image bytes to
the new image (without overwritting the new file)
$ losetup -o 32256 /dev/loop0 new_image # /root doesn't actually live
at front of image for HVM
$ fsck -f /dev/loop0 #check the new image for errors
$ echo "new_image_bytes / 8225280" | bc # this is number of cylinders
in new device
$ fdisk new_image
$ # go to the "extra" menu and type "c" and type the new number of cylinders
$ # go back to main menu
$ # delete primary partition (yes, delete it)
$ # add new primary partition w/ defaults # resize partition to size of
actual image file
$ # 'w' write out fdisk and exit
$ resize2fs /dev/loop0 20G #resize ext3 to whatever GB you want MINUS
a gigabyte ..... just to be on the safe side
$ losetup -d /dev/loop0 # get rid of the loop device
$ edit xen config and test new image (don't delete the old one!)

wget example

The following script
convert windows accept filenames
convert links to local link
convert cgi/jsp link to html file name
recursive download 3 levels

wget  --restrict-file-names=windows -E -r  -p  -l 3  -k   -Ahtml -Ahtm -Ajpg -np
 -nv -nd -nH http://www.fda.gov/Safety/Recalls/ArchiveRecalls/2007/default.htm

The following example want to download only html file from
one website and put it all the files in one certain directory
and convert link relationship to make it cross-referable.

#!/usr/bin/env bash

MONTH=("jan" "feb" "mar" "apr" "may" "jun" "jul" "aug" "sep" "oct" "nov" "dec")
YEAR=("00" "01" "02" "03" "04" "05" "06" "07" "08" "09" "90" "91" "92" "93" "94"
 "95" "96" "97" "98" "99")


for mon in ${MONTH[*]}
do
    for year in ${YEAR[*]}
    do
        x=$mon$year
 mkdir ~/fay/$x
        cd ~/fay/
 wget  -p -r -l 2  -k -nc -Ahtml  -P$x  -np -nv -nd -nH  http://www.cpsc.
gov/cpscpub/prerel/prerel$x.html
    done
done

echo "Wj Script done"

wget useful options

#usefule options

To download a single page and all its requisites (even if they exist
on separate websites), and make sure the lot displays properly
locally, this author likes to use a few options in addition to -p:

        wget -E -H -k -K -p http:///

-np     no parent
-H
       --span-hosts
           Enable spanning across hosts when doing recursive retrieving.

-A.c download .c file
-E   add html extension to cgi/jsp script links
-K
       --backup-converted
           When converting a file, back up the original version with a .orig
           suffix.  Affects the behavior of -N.
-k     
       --convert-links   The same as the above, but convert the links in the HTML
         files to point to local files, so you can view the
         documents off-line:

                 wget --convert-links -r http://www.gnu.org/ -o gnulog
---restrict-file-names=mode
           windows doesn't accept certain file names like <>?
           it can replace with certain chars like ? to #

-nc
    --noclobber(no overwrite) 
           With -nc, download the same file will overwritten the old. 
           With -r -nc, download the same file will avoid overwriting, just use the original copy. 
           When running Wget without -N, -nc, -r, or p, downloading the same
           file in the same directory will result in the original copy of file
           being preserved and the second copy being named file.1.
 -F
       --force-html
           When input is read from a file, force it to be treated as an HTML
           file.  This enables you to retrieve relative links from existing
           HTML files on your local disk, by adding some url to
           HTML, or using the --base command-line option.
-p
       --page-requisites
           This option causes Wget to download all the files that are
           necessary to properly display a given HTML page.  This includes
           such things as inlined images, sounds, and referenced stylesheets.

 -q
       --quiet
           Turn off Wget’s output.

 -nv
       --no-verbose
           Turn off verbose without being completely quiet (use -q for that),
           which means that error messages and basic information still get
           printed.

wget -r -p -l 2 -np -nc -q  http://www.cpsc.gov/cpscpub/prerel/prerel.html

12球称重问题

脑筋问题挺好玩的。

有12个乒乓球,其中有一个重量不一样,但不知是轻是重,
请用天平称三次,把坏球找出来解法
----------------------------------

A1A2A3A4
B1B2B3B4
C1C2C3C4


A1-4 vs B1-4

• 如果A=B

那坏球在C堆里

C12 vs A12，如果相等，坏球在C34里，C3 vs A1 ....
如果不等，坏球在C12里，C1 vs A1 ...

• 如果A<B(A>B同样处理，switch AB即可)

A1B1C1 vs A2B2B3

1. A1B1C1 = A2B2B3

A3A4, B4, A3B4 vs C1C2

• A3B4 vs C1C2,
• if 相等，==》A4轻

• if < ==》A3 轻
  
• > ==》B4重

2. A1B1C1 < A2B2B3
不可能B1轻或者A2重，所以或者A1 轻，或者是B2B3重，

• A1B2 vs C1C2
  

3.A1B1C1 > A2B2B3
either B1 重或者A2轻，

• B1 vs C1 ....

思路关键就是要记住第一步的轻重对解决问题有指导性作用。

-----------------------------
还有8球问题，比较简单，分成332，先比较33就可以了。
这一题好像要知道坏球是重还是轻才行。

Multikernel

Cache coherence become a problem with the increase of cores.

Using message passing instead of shared memory can solve the problem.

http://docs.google.com/present/view?id=dc3nqpnt_19hsg376hs

try to explain slide 21,

Broadcast vs Multicast:

TLB shootdown – the process of maintaining TLB consistency
by invalidating entries when pages are unmapped
– is one of the simplest operations in a multiprocessor

OS that requires global coordination. It is also
a short, but latency-critical, operation and so represents
a worst-case comparison for a multikernel.

Broadcast performs badly due to the cache-coherence protocol used
by AMD64 processors. When the master updates the line, it is invalidated
in all other caches. Each slave core then pulls the new copy from the master’s
cache.

Multicast protocol, the master sends a URPC message to the first core of each processor,

which forwards it to the other three cores in the package.
Since they all share an L3 cache, this second message is much cheaper.

IP Loopback

Table 4 shows that Barrelfish achieves higher throughput,
fewer cache misses, and lower interconnect utilization,
particularly in the reverse direction from sink to
source. This occurs because sending packets as URPC
messages avoids any shared-memory other than the
URPC channel and packet payload; conversely, Linux
causes more cache-coherence traffic for shared-memory
synchronization.
Barrelfish also benefits by avoiding kernel crossings.

AICT new problem

I got into two problems right now with AICT
AICT Performance not as good as expected.

1.

S CPU1 T CPU2
I get the same performance(7300tran/s) with raise_softirq to T.
with aict I cannot get any performance improvement.

Since at CPU 2 side , T with C1 and C2 are competing for
CPU resources, with aict, T should get some kind of priority
hence smaller


2.
If I give up CPU at Target side right after target finish its I/O
shedule() doesn't take effect as I expected.
Problem: xenloop doesn't know when target should give up
cpu during xmit_packets().

Task: to find after schedule(), when could S get schduled