KERNEL NEWS ZACK’S KERNEL NEWS

Code Removal Adrian has also tried removing the New Features The effort to get bad, unmaintained, or VIDEO_ZR36120 driver, but in this case, With all the old code coming out of the unused code out of the kernel continues he’s encountered some resistance. It kernel, there are still a lot of features unabated. JFFS is slated for removal in turns out Joe Feise uses the driver and that are being added and old features 2.6.21, unless some secret cache of users has even patched it to work on his home being fixed. Andries Brouwer recently is discovered. It’s been unmaintained for system; though when he submitted his had to read a Minix version 3 filesystem, years and been superseded by JFFS2 for patch to the official maintainer, Pauline only to discover that 2.6.19 didn’t sup- years as well. Jeff Garzik made the pro- Middelink, he got no reply and the patch port it and neither did the latest 2.6.20 posal to get rid of it and submitted a was not adopted into the kernel. release candidate. patch; and Andrew Morton has no objec- Peter Schlaf also uses the driver and A quick search revealed that Daniel tion, so it looks like very nearly a done had been using a forked version of Pau- Aragones had already written a patch to deal. Farewell JFFS! line’s code to support it. On the mailing support the filesystem, and after clean- Getting rid of OSS sound drivers has list, Schlaf offered to take over maintain- ing it up quite a bit, Andries submitted been on Adrian Bunk’s to-do list for ership of the driver, and Pauline said this it to the kernel list for inclusion. Daniel years, and he’s gotten rid of quite a few would be fine. And Mauro Carvalho approved of Andries’s changes, and of them, as more and more ALSA drivers Chehab offered to accept any patches Andrew Morton said the patch looked start working just as well for the same coming in for the driver, since he main- harmless enough to accept right away. hardware. DMASOUND_PMAC and tains the V4L subsystem as a whole. So Jiri Kosina has resurrected the ipwire- SOUND_ES1371 are the latest two head- it looks as though this particular driver less_cs 3G PCMCIA network driver, orig- ing for the chopping block. Thomas has been saved from death at the very inally developed by Symmetric Systems, Sailer and Kyle Moffett have both given last moment. and forward-ported it to the current 2.6 their personal confirmation that these The VIDEO_PLANB driver also found kernel. The original authors are also par- drivers may be safely removed, and no a new maintainer. Adrian posted a patch ticipating in debugging and enhancing voices have been heard in objection. to remove it, but Benjamin Herren- the code. There is a lot to be done! The schmidt volunteered to take a stab at fix- code itself violates CodingStyle rules and The kernel ing it up. And Michel Lanners, the origi- fails to use various in-kernel resources; mailing list com- nal author of the code, expressed great and its support of the hardware itself is prises the core of interest in helping out with explanations buggy. But Jiri and the rest have felt a Linux development of the hardware and existing code, al- pulse on this one! A working driver can- though he had no time to actually main- not be too far in the future. activities. Traffic vol- tain the driver. Sascha Sommer has implemented an umes are immense, The SCSI_AMIGA7XX driver is up in experimental driver for the Ricoh SD often reaching ten the air now. Adrian posted a patch to re- Card reader, as found in notebooks such thousand messages move it, but Geert Uytterhoeven offered as the Samsung P35 and the DELL X300. in a given week, and his assessment, saying, “There’s a fix Samuel Thibault is super excited to see keeping up to date available to convert this driver to the this work going on, and Pierre Ossman with the entire scope of development new 53c700 core. But it needs the new offered his encouragement as well. The is a virtually impossible task for one DMA framework, which still causes a driver currently offers read-only support, person. One of the few brave souls to few regressions on m68k that are being is slow, and has a variety of other prob- take on this task is Zack Brown. worked on.” Whether this will be lems; but Samuel was able to use it on Our regular monthly column keeps enough to keep the driver alive is an his X300. you abreast of the latest discussions open question. David Brownell has written a driver and decisions, selected and summa- Other drivers are more cut and dry. framework for the real-time clock built rized by Zack. Zack has been publish- Adrian submitted patches to remove the into PCs and some other platforms. This ing a weekly online digest, the Kernel BINFMT_IRIX driver, the FB_S3TRIO is not an entirely new thing, a real-time Traffic newsletter for over five years driver, the SUN_AURORA driver, and clock driver exists already. But David’s now. Even reading Kernel Traffic alone the OAKNET driver. No one stepped up work has a more extensible feature set, can be a time consuming task. to claim them. In the case of the S3Trio, and a standard user interface via . Linux Magazine now provides you Geert Uytterhoeven pointed out that a Drawbacks are that it’s only been tested with the quintessence of new generic S3Trio driver was in the on x86 with ACPI and no HPET and activities, straight from the horse’s works, so there was really no need to therefore can look forward to some mouth. cling to FB_S3TRIO. “growing pains,” as David puts it.

14 ISSUE 77 APRIL 2007 KERNEL NEWS

Adver tisement Philip Langdale has coded Maintainership up some support for the up- Richard Knutsson suggested coming SDHC (Secure Digital changing the format of the High Capacity) flash cards. MAINTAINERS file slightly to John Gilmore donated the include the actual configura- hardware, and as the SD Card tion variable for each project Association has published listed there. This way, anyone useful specifications, he was having a problem with a able do it! The code has not driver could run a script that been widely tested yet, but would query the MAINTAIN- Philip has not lost any data ERS file for the correct owner in his recent tests. and project status. Several David Lopez recently sub- kernel folks support this idea, mitted a driver to support and Matthias Schniedermeyer LabJack U3 and UE9 USB also suggested splitting the data acquisition devices. MAINTAINERS file into These devices provide a smaller files, as was done general purpose connection with the monolithic kernel between a computer and configuration file before laboratory instruments. KConfig came along. Among Nicolas Ferre has modified folks like Andrew Morton, the ads7846 driver code to however, these ideas have support the ads7843 touch- not yet taken root. As An- screen controller. drew puts it, “I find that the It seems there is also some most practical way to find out work being done on that who really maintains a driver driver elsewhere, and folks is to run git-whatchanged on seem to be scrambling it and see who has been around trying to sort out doing stuff to it.” which patches are where, and Meanwhile Leonard Nor- who to ask. But there don’t rgard has also been interested seem to be any insurmount- in automatically parsing the able problems. MAINTAINERS file; and to that end, he has submitted a Rooting Out Spam patch to convert occurrences At vger, Sort Of of “Orphaned” status lines Matti Aarnio is trying Post- into the more canonical “Or- grey filtering for the linux phan” state. It remains to be kernel mailing list. This in- seen, whether scripts that volves delaying each email by read the MAINTAINERS file up to five minutes, because will soon be replaced by git- legitimate mail servers will whatchanged. automatically try to resend Meanwhile, there have the email, while most spam- been some actual maintainer- mers don’t bother. ship changes. Aside from the Matti’s initial measure- odd updated email address or ments show a 90% drop in website URL in various proj- spam coming into the list as ects, Martin Waitz has abdi- a result of this change. How- cated maintainership of the ever, he predicts that within kernel-doc DocBook effort, 200 days, most spammers leaving the project entirely in will wise up and start retry- the hands of Randy Dunlap, ing their deliveries like legiti- his comaintainer. Martin said mate mail servers. So his he didn’t have the time to put change will only provide into this effort anymore, so it a brief respite from spam, made no sense to keep him- rather than a long-term self listed as an official main- improvement. tainer.

W W W. L I N U X - M A G A Z I N E . C O M ISSUE 77 APRIL 2007 15