Letters

Please congratulate Lydia Kinata for the magazine's new design. I have been waiting a long time for this kind of improvement. Best regards and keep improving.

—Jorge Carminati

I finally had time to read the September issue. The first thing I noticed was the new design. Looks great, keep the change! I've had a subscription for over two years and this, to me, was the best issue (look and content) I've received.

—Matt C.

Robin, I got a copy of the Linux Journal at LinuxWorld and read your article on DreamWorks—what a great visit that must have been!

Given what you saw there and may already know of, I was wondering if you could offer a suggestion. I have a teenage son that is very interested in animation, but the applications to do that are pretty expensive. Can you suggest some Linux-based rendering applications that are more affordable?

—Jack C.

Robin replies: Jack, we had a great time at DreamWorks. But DreamWorks is just the first. All the studios have Linux migration projects underway.

You may want to start with Blender, a popular, freely distributed 3-D modeling package for animation, rendering, interactive 3-D and game creation. Unlike most no-cost animation tools, there are printed books available on Blender. Other tools that may be of interest include K-3D, Ayam, AC3D, Flow, Radiance and BMRT. You can find their web pages by searching at Freshmeat.net

In the September 2001 LJ article “Loadable Kernel Module Exploit”, I was interested in the technique for hijacking the sys_call_table pointer and wrapping it within your own function. I used to do something very similar in DOS in the good ol' days, writing TSRs (Terminate and Stay Resident) and other programs that “borrowed” the bios or dos interrupts vectors.

For programs, it was a simple case of restoring the original interrupt vector (pointer), but with TSRs, one had to be careful before restoring the original pointer. The problem being that if several TSR utilities were installed, two might hijack the same interrupt (often the keyboard). A properly written TSR would, before uninstalling, check that the current pointer pointed indeed to itself, if it didn't, then it could not uninstall as by doing so it would clobber another, later installed, TSR.

The cleanup_module(void) routine in the article performs no such check before restoring the sys_call_table[SYS_write] pointer, something which at first sight appears dangerous to me in a multitasking environment, especially one where the kernel can load and unload modules automatically.

—John Jacq