I would deprecate them now. From a user's perspective, there are just too many axis simulations, it becomes very confusing when looking to try a feature. Thats what I find anyway. Rod Webster
On 2026-09-08 14:59, Luca Toniolo <[email protected]> wrote: > On 9/8/2026 12:01 AM, andy pugh wrote: > > I think that most of these would actually build with preempt-rt, but > > are not built as they can't be tested. > > (Ie, I think that it is a testing problem, not a building problem, > > that has them omitted from the uspace builds) > > I tested that assumption: with uspace flags, none of the seven C drivers > compile as they stand: > > - hal_stg, pci_8255, hal_ax5214h, hal_vti: asm/io.h, no userspace equivalent > > - hal_motenc, opto_ac5: pci_get_device, ioremap_nocache, raw <linux/pci.h> > > - hal_evoreg: closest, uses rtapi_io.h, but still calls ioremap/writew > directly > > - pcl720.comp: asm/io.h plus kernel types > > Only pluto_servo/pluto_step compile today (userspace sys/io.h path since 2007) > > So it is a build problem too, not only testing. The porting path exists > (rtapi_pci.h/rtapi_io.h shims, hm2_pci and hal_gm show how), but each > driver needs real work plus hardware to validate. > > That leaves the direction question you didn't take a position on: > > A. Port and keep maintained, on the assumption users exist but haven't > shown up yet. Cost: porting effort per driver, review bandwidth, and > code that ships untested anyway until a tester appears. > > B. Deprecate now, bring back on request. Add the deprecation note, > announce on forum + mailing list, remove after one cycle. If a real user > appears later for a specific card, the git history keeps the driver and > porting it then is the same effort as porting it now, except then we > have a tester from day one. > > I favor B. Shipping freshly ported drivers with zero hardware validation > is the same untested state we have today, just with more code to > maintain. Do you see a concrete reason to prefer A for any specific > card?PS: Forum poll result: a pci_8255 user reported in on the forum. He is > running a 3.x RTAI kernel, so likely an old LinuxCNC release that will > keep working regardless of what master does. Unknown yet whether he > upgrades or would test a uspace port; forum went down before I could > follow up. Will update when I can reach him. > > This is the distinction the poll needs to draw: "card still in service" > vs "card owner plans to upgrade past the removal". Only the second > argues for keeping or porting. > > _______________________________________________ > Emc-developers mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-developers _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
