ls rise lab

Some time back, I decided that I will go back to my room. But then what will I do there ? Watch another movie or some music channel. No no. Already people are complaining that I am watching Television a lot, that too music channels :P. Instead, I decided to write this blog entry, which I wanted to write from a long time. This time I will also be technical. Always philosophical or funny is not good. Sometimes it is good to play with the devil hiding in the details ;)

Now I am going to write about people I know in rise lab. I virtually lived in this lab for more than 2 months. I lived there because:

1. There are many good people there.

2. I can access internet all the time.

3. It is very peaceful in the night and there are nice chairs which are suitable for sleeping.

Don't ask me why I didn't stay in systems lab where all my friends are there. It was a run-away mechanism from somebody. Only Shiny knows the real truth. Even I forgot :P.

Note: My blog, Picasa pics, etc. are all personal to me. I include all my friends and relatives details in them. But that doesn't mean that I am making them public or advertising them. Pics are never made public. They are shared with the right persons. Blog is public, but everybody knows how famous it is :P The main purpose of all these is to capture few memories of happy moments in pics and some worthy thinking in blog. One should remember that the artificial facts like sexy, handsome, etc. can never describe a real person. They are just artificial and they are not a measure of how good their heart is.

First comes Balvir Kumar. I first met him through my granny (check the note in the end to know about this granny). But then I forgot him. Later when I started living in rise lab, I always sat beside him unknowingly. As usual, I peep through and watch what others are doing :) My intrusive nature :P. So I saw him struggling with some latexpdf thing and immediately took over his computer and fixed it :D. That's all. From next day he started talking to me :) ha ha.. Karthik raghavan explained me a bit about his project. So I inquired about his project. It is very interesting ! He is also very enthusiastic person and always wanted to put fundaes on his project. But I postponed many times due to lack of time. Finally one day before I left insti, I asked him if he can still put fundaes. He said that he'll come to the lab the next day and put fundaes. Awesome ! he'll come to lab just for me and put fundaes. I know his house is near and he has bike :P. But still he is god. His project is very good.

About his project: "Handling Cache Faults by Dynamic Binary Rewriting". Scared by this big title ?. Don't worry, even I copied it from his thesis paper I have now in my hand. Few points regarding his project:

a. It is about virtually indexed caches.

b. Assumption: A cache set can become faulty. Therefore the region mapped by that set in VM is faulty, ie it is not accessible.

c. Solution: don't use that region. How ? if it is data, map it to some other location. If it is code move it some location and execute there.

d. The solution above is simple said, but there are many issues.

e. Data cache faults are handled by segmentation. The faulty region is not mapped and the data in that region is relocated to a new segment. Therefore accessing faulty area will give segfault, which will reroute the data fetch to the new segment. Still there are other issues with boundaries.

f. Instruction cache faults are also handled in similar fashion. The code in the faulty area is moved to new segment and a jmp is put in the code before the faulty area. Now the problem comes with offsets. The jmp/call instructions with fixed offsets are dynamically rewritten for the new locations. But for indirect control transfer like jmp (%eax), it is not possible to rewrite the instruction and execute happily. Instead a int 3 is put in that location and the debug routine will execute the original instruction. If you know Intel arch, a fault will rerun the instruction. All these complications are also handled.

Coming back to non-technical side again, after discussing about the project we talked all arbitrary things. I asked him where he works and about his work. It is also very interesting. He works in defense labs. There they make custom boards and run custom apps without OS on them. So his work is to write those standalone apps on those boards. It is vvv interesting. Just that it is a government job and not paid so highly as corporate ones. His friends are working abroad in firms. During his time, the jobs were not so abundant and he got this job. He liked it very much and he stayed with it. He recently got married too and his wife wants to go home :) . With people like me, nothing stays secret :P. So he is going home this Friday and he is not going to come back soon :'( He is also not coming for convocation. He gave me his Chennai phone number and also Chandigarh one. So I told him that I will come to Chandigarh someday to meet him :D

Similarly other good people who are leaving this time are Girish, Karthik Raghavan (music lover like me :P) and Arun Tejasvi. I can write big stories about Girish and Karthik, but I don't have so much enthu now. I may write later. Again I am mentioning these people because they are leaving this summer and I should not forget them.

There are many other people who are still going to stay in insti like pawan, abhinav and others. I will write about them when they leave ;)

Feeling so tired now. I had one cup of horlicks. But nothing seems to work. Very tired. One last thing. Today I found the reason why my cointerns are calling me garimella. According to north indian style, if we have 3 names like garimella jagannadh kashyap, the middle name will be father's name; the last one is surname and the first one is my name. Similarly sachin ramesh tendulkar, sumit ramesh somani (my cointern from Jaipur).

Note: granny, mentioned in the 6th para, is an adopted granny. She is young secret person, whose details I can never reveal.