From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; rv:1.7.3) Gecko/20041020 Firefox/0.10.1 Description of problem: On a video card where kudzu (or whoever) detects less than 1MB of video RAM (such as a Diamond FireGL 1000 Pro), ddcprobe claims the card has 0 MB of RAM. This is misleading. Version-Release number of selected component (if applicable): rhpl-0.148-1 How reproducible: Always Steps to Reproduce: 1. Set up a machine with a Diamond FireGL 1000 Pro, or with another card that pretends (from kudzu's POV) to have less than 1MB of RAM, or with a card that actually has less than 1MB of RAM. (I know this is impractical for Red Hat to do, but you *do* want steps for reproducing this bug, right?) 2. Install a Red Hat, FC, or RHEL distribution from (I think) the last several years. (FC1 or later will definitely suffice.) 3. As root, run /usr/sbin/ddcprobe Actual Results: Memory (MB): 0 Expected Results: One of the following: Memory (KB): 512 Memory: 0MB (512KB) Memory: 512KB Memory (MB): 0 Memory (KB): 512 or *at least* (although this is ugly IMO): Memory (MB): 0.5 Additional info: I was going to write a patch, but then I realized that I'm not sure how the output should be formatted. I might write and attach a patch anyway (against FC3) though.
Created attachment 106276 [details] ddcprobe-stub patch to fix this bug This patch still displays in MB if the amount of video RAM is divisible by 1024K, and displays in KB otherwise. I've tested it on both a Diamond FireGL 1000 Pro (now it tells me 512KB) and a 16MB nVidia Riva TNT2 (output is unchanged, it still says 16MB).
Basically committed to CVS -- changed the test to be if mem < 1024 so that cards that do something weird like report 64.5 MB would still report 64 (not that I know of any of those, but still)
Sounds good to me. Thanks!