Josef “Jeff” Sipek

Whee! This weekend happened to be filled with coding.

First, I realized that it’s been 15 months since the last HVF release. I looked at the list of commits since then, and it was a sizable enough list to warrant a new release. Since there wasn’t a whole lot “Oh my! Must have!” it ended up being just another release candidate.

Once released, I looked around my patch directories to see what else I should hack at. The installer patch caught my eye.

I had a branch for the DASD loader work for a while — and the loader is complete. On the same branch, I had uncommitted code to implement a simple installer program. I started working on it almost a year ago (at least that’s what I gather from the bug report:  146), but I kept all the code uncommitted. I did some simple cleanup, and committed the work-in-progress code. Then, over the next few hours, I managed to get a very large portion of it done. The last piece that needs to be implemented is the EDF handling. That is, the code that lets HVF use  CMS file system for config files, etc.

In other news, I ran Doxygen on the HVF codebase. The output wasn’t as impressive as I hoped it would be. Part of it is probably because there are way too many functions that aren’t documented  76. This would be a good starter project for anyone looking to start hacking on HVF. Since I’m on the topic of looking for help, I realized that the HVF website is awful and could use some help — both look & feel as well content.

Lastly, I’d like to post a link here to the HVF Ohloh page.

Over the past few months, I have pondered many aspects of C++. I thought it would be good to write my concerns and conclusions down somewhere. That somewhere ended up being right here — my blahg.

I was thinking about C++ and how quirky it is. And then I considered the scenario where you dynamically allocate an object, but the constructor throws an exception. Would that free the allocated object? Well, section 14.4.4 in the C++ book answers that:

What happens if X’s constructor throws an exception? Is the memory allocated by the operator new() freed? For the ordinary case, the answer is yes…

When the placement syntax is used, the answer cannot be that simple. Some uses of that syntax allocate memory, which then ought to be released; however, some don’t. Furthermore, the point of using the placement syntax is to achieve nonstandard allocation, so nonstandard freeing is typically required. Consequently, the action taken depends on the allocator used. If an allocator Z::operator new() is used, Z::operator delete() is invoked if it exists; otherwise, no deallocation is attempted.

Curse you, Bjork Stroustrup!

I was just looking at the source for Postfix, when I came across this function:

/* mail_conf_read - read global configuration file */

void    mail_conf_read(void)
{
    mail_conf_suck();
    mail_params_init();
}

It turns out that mail_conf_suck reads in the config file, and then mail_params_init does all the dirty work of initializing the internal data structures based on the config.

Anyway, that’s the random thought of the day. I found it marginally amusing.

Edit: the code in question is in src/global/mail_conf.c.

Earlier this year, I heard of a company that tried to make a product out of dynamic binary translation of x86 code to  z/Architecture. Recently, I decided to look at what they do.

The company is called Mantissa Corporation, and their binary translation product is called z/VOS.

Much like VMWARE, they cache the translated code, in z/VOS’s case it’s really a must otherwise I’d guess the cost of traslation would make the result unusable. I like how they used VNC (see the demo mentioned below) to give the virtual x86 box a display.

There is an official blog that has some interesting bits of information. For example, they hint at how they use multiple address spaces to give a the x86 code the illusion of virtual memory. I am not quite sure why they list  Decimal Floating Point facility as a requirement. Unfortunately, it has been a few months since the last update.

Their website also happens to have a demo of a small x86 assembly operating system starting up and running under z/VOS. I find this fascinating.

Recently, I wanted to figure out whether or not an application I was analyzing was memory bound or not. While on this quest, I was introduced to Performance Application Programming Interface (PAPI).

There is a rather good HOWTO that shows step-by-step instructions on getting it all running on Debian. The text below is more or less just a short version of that HOWTO, with my thoughts interspersed.

PAPI is a library that hooks into the hardware performance counters, and presents them in a uniform way. Installation is rather simple if you pay attention to the installation instructions.

1. Get the kernel source
2. Get the perfctr tarball
3. Extract the sources, and run the update-kernel script. I really mean this, if you try to be clever and apply the patch by hand, you’ll have a broken source tree. (The script runs patch to fixup some existing kernel files, and then it copies a whole bunch of other files into kernel tree.)
4. Configure, build, install, and reboot into the new kernel
5. You can modprobe perfctr and see spew in dmesg

That’s it for perfctr. Now PAPI itself…

1. Get & extract the source
2. ./configure, make, make fulltest, make install-all

That’s it for PAPI. The make fulltest will run the tests. Chances are that they will all either pass or all fail. If they fail, then something is wrong (probably with perfctr). If they pass, then you are all set.

There are some examples in the src/examples directory. Those should get you started with using PAPI. It takes about 100 lines of C to get an arbitrary counter going.

Some other time, I’ll talk more about PAPI, and how I used it in my experiments.

Alright, it ain’t rocket science. When you are trying to decide which filesystem to use, and you see a 7 year old article which talks about people having problems with the fs on Red Hat 7.x (running 2.4.18 kernels), are you going to assume that nothing changed? What if all the developers tell you that things changed? Are you still going to believe the slashdot article? Grrr… No one is forcing you to use this filesystem, so if you believe a 7-year old /. article, then go away and don’t waste the developers’ & others’ time.

Insanity! Someone has made it possible to write kernel modules in Haskell. (FYI, Haskell is a functional language with very strong typing.) Currently, they support only x86, but I wouldn’t be surprised if some other architectures got a port soonish.

I just remembered a fun fact about the  Apollo Guidance Computer (AGC).

A few years ago, I was tinkering with emulating one. It kind of worked, but that’s not important.

The AGC was an accumulator architecture, with 15-bit words (the accumulator (A) and few other “registers” were 16 bits), and 1’s complement arithmetic.

Now, the fun fact. There was an instruction called ’CCS’. It took an address, loaded the accumulator with the value at the address, and then performed a 4-way branch. The easiest way to explain it is with a some code that demonstrates what happened with some C-style pseudocode (A = accumulator, Z = program counter):

Z = Z + 1;
A = mem[operand];

switch(A) {
        case POSITIVE:
        	A = A - 1;
        	/* Z is already incremented */
        	break;
        case POS_ZERO:
        	/* A is already a zero */
        	Z = Z + 1;
        	break;
        case NEGATIVE:
        	A = (~A) - 1;
        	Z = Z + 2;
        	break;
        case NEG_ZERO:
        	/* A is already a zero */
        	Z = Z + 3;
        	break;
}

So, in a program, you could see something like:

CCS addr1
TC  addr2
TC  addr3
TC  addr4
TC  addr5

So, data from addr1 was loaded, and then one of the TC instructions (TC = Transfer Control = a branch instruction) was jumped to depending on the value, then when the TC got executed, it unconditionally branched to some other address.

Of course, you didn’t have to use TC, you could use any valid instruction and CCS would happily jump to it.

My understanding is that sometimes when CCS was used, some of the 4 possible targets were impossible. Not to waste memory, a committee was organized to keep track of these holes, and fill them in with useful constants.

Earlier today, someone I know sent me this Time article. I started reading the article, but something seemed a bit odd. To not spoil it for you, here’s the text of the article:

The computer has become a main stay of big business in the U.S., but most small and medium-sized companies still find it too expensive for normal use. Last week two of the biggest computer makers, General Electric and Control Data Corp., introduced new systems that will offer the small business man the same computer advantages as the biggest corporation. Their move to what is called “time sharing” is part of a growing trend to market the computer’s abilities much as a utility sells light or gas.

Dial for the Answer. Business some time ago began using computer centers to process data cards, count receipts or keep track of airline reservations from distant offices. Time sharing goes much beyond that. It links up as many as 500 widely separated customers with one large computer, lets each feed its own problems to the machine by telephone through a simple typewriter console. The time-sharing computer can answer questions in microseconds, is able to shift back and forth swiftly among the diverse programming needs of many companies, small and large.

Although still in its infancy, time sharing is already being used by business, government and universities. Boston’s Raytheon Co. prepares contract proposals, and Arthur D. Little solves problems in applied mechanics through a time-sharing system run by Cambridge’s Bolt Beranek & Newman. An other time-sharing firm, Keydata, will soon take up the problems of Boston distributors of liquor, books, automobile parts and building materials. Control Data, which introduced two time-shared computers last week, will open the U.S.’s biggest sharing center in Los Angeles next year. General Electric already has 88 customers, last week added a New York center to its service centers in Phoenix and Valley Forge, Pa.

From New York, IBM gives shared-time services to 50 customers, including Union Carbide and the Bank of California. Under G.E.’s system, a company can rent the big G.E. 265 for 25 shared hours a month for only $350, compared with a normal monthly rent of $13,000 for individual computers.

Plugging Them In. Some companies have discovered that time sharing has reduced to one-fiftieth the time needed to answer a problem, have found access to a large computer more profitable than ownership of a small or medium-sized machine. The Massachusetts Institute of Technology, one of the pioneers in time sharing, now has 400 users for its IBM 7094 computer, has served scientists as far away as Norway and Argentina. Experts predict that by 1970 time sharing will account for at least half of an estimated $5 billion computer business, will be used as widely and easily as the telephone switchboard.

Yep, that’s right, this article is dated: Friday, Nov. 12, 1965. :)

Mendel Rosenblum, a Stanford University associate professor, VMWare co-founder, became an ACM Fellow, in 2008. The page about his fellow citation reads:

For contributions to reinventing virtual machines.

Very amusing.