Re: [Machinekit] Mesa 6i25 & 7i76 - Operation not permitted with new Machinekit Version

[schoone...@gmail.com]

Well I can see the cause of all this, but the solution without generating a warning somewhere in the package builds is far from simple. When this quite low level C was written, warnings were not as numerous as nowadays versions of gcc generate and probably not top priority anyway so long as it worked. I spent quite a long time trying various ways of removing warnings, but none were satisfactory for all usages of the variable, across all architectures. It essentially comes down to 32 bit and 64 bit differences in data type size. If you then specify a format size in a printf operation, it will always generate a warning under one architecture or another. Assigning with a (void*) cast, will do so too, hence the making of L1 and L2 void * probably The idea of the `address of` reference &L1 was probably to prevent a warning regards printf format %p requiring void **, but I would think it does not return what the function wants. In any case I have reverted the commits and removed the -Werror CFLAG for now to enable packages to build. Will take a while to filter through, not been able to test on my Mesa carded machine due to other issues, hence reverted in its entirety. Just using a package prior to these commits will keep you going for now. Thanks for the report and your testing to quickly narrow it down. regards   On 05/03/19 19:24, Dennis wrote: Digging through github history in between 4th and 9th of December 2018 I found a commit (https://github.com/machinekit/machinekit/commit/e207745f52181562d22cacd636bc03721d2c2587) that modified the function pci_enable_device in rtapi_pci.c. This is the same function that throws the parse error in my linuxcnc.log.  Maybe this is the problem... Am Dienstag, 5. März 2019 20:13:23 UTC+1 schrieb Dennis: By geometric search I could narrow down the first 'problematic' package, I hope this is helpful to you: the last working version is: 0.1.1543935482 the first non working version is: 0.1.1544363499 Thanks again for your help! Best regards Dennis Am Dienstag, 5. März 2019 18:56:37 UTC+1 schrieb Schooner: No need for the github issue, we are looking at it! The commit that was OK was quite old, it would be good to try and narrow down a bit to where the problem arose. It will not have been very recently, since those were all ARM config changes. If you are happy to do it, I can only suggest picking a package from a mid-point, say the first one in 2019 and installing that. If it fails work back to say mid Dec 2018, if it succeeds work forward to end of Jan 2019 and so on. In the absence of specific changes to hm2_pci, I suspect that another change, of which there were quite a few to correct warnings etc, must have had an unforeseen effect elsewhere. Anything you can do to isolate a commit or series of commits as causing this, would be very helpful Tomorrow hopefully I can test on a 5i25/7i76 setup in the workshop, so having an idea where to look would speed things greatly regards On 05/03/19 17:08, Dennis wrote: Thanks to you both! I've done some further testing: 1) Mesa 6i25 in different PCI-E slot --> same result, not working 2) Different Mesa 6i25 from another working linuxcnc PC in original slot and second slot --> not working 3) Procedure from Schooner (downgrade to last working version) --> !working! 4) To countercheck I upgraded again to latest version of machinekit, the one that was not working --> not working So to me something with the new version of machinekit is now working with my Mesa 6i25. It is not the card, not the PC and not the configuration. Should I file an issue on the github tracker? Best regards Dennis Am Dienstag, 5. März 2019 13:31:57 UTC+1 schrieb Schooner: On 05/03/19 11:32, Bas de Bruijn wrote: Not sure how much i can help here. Further on there’s this section: 8888:rt halg_xinitfv:90 HAL: initializing component 'hm2_pci' type=1 arg1=0 arg2=0/0x0 Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt PCI_ID: 2718:5125 2718:5125 Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt RTAPI_PCI: DeviceID: 2718 5125 2718 5125 Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt RTAPI_PCI: Calling driver probe function Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt RTAPI_PCI: Enabling Device 0000:03:00.0 Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt Resource 0: 0xf7e00000 0xf7e0ffff 00000000 Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt Failed to parse "/sys/devices/pci0000:00/0000:00:1c.2/0000:02:00.0/0000:03:00.0/resource" Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt hm2_pci: skipping AnyIO board at 0000:03:00.0, failed to enable PCI device Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt Driver probe function failed! Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib:8888:rt hm2_pci: error registering PCI driver Mar  5 11:38:33 labor-linuxcnc-m2 msgd:0: hal_lib: Did other things get updates too perhaps? hm2_pci has not changed for 3 years and I cannot immediately see any other changes that might affect First step is to reverse the process apt remove machinekit* Then download http://deb.machinekit.io/debian/pool/main/m/machinekit/machinekit_0.1.1543327459.gite758f69-1~stretch_amd64.deb http://deb.machinekit.io/debian/pool/main/m/machinekit/machinekit-rt-preempt-dbgsym_0.1.1543327459.gite758f69-1~stretch_amd64.deb and install with dpkg -i machinekit_0.1.1543327459.gite758f69-1~stretch_amd64.deb machinekit-rt-preempt-dbgsym_0.1.1543327459.gite758f69-1~stretch_amd64.deb from wherever you downloaded them to Then try again with DEBUG=5 set and attach the linuxcnc.log whatever the result. (blank linuxcnc.log first) On the face of it the error is failing to contact the board, not failing to load the driver and the error from the driver is resultant from that. If you did not have this working immediately before ( the same day) you updated machinekit, I would check all cabling and possibly remove the 6i25, clean the slot and board contacts with methylated spirits or similar, re-seat and try again. -- website: http://www.machinekit.io blog: http://blog.machinekit.io github: https://github.com/machinekit --- You received this message because you are subscribed to the Google Groups "Machinekit" group. To unsubscribe from this group and stop receiving emails from it, send an email to machinekit+...@googlegroups.com. Visit this group at https://groups.google.com/group/machinekit. For more options, visit https://groups.google.com/d/optout. -- website: http://www.machinekit.io blog: http://blog.machinekit.io github: https://github.com/machinekit --- You received this message because you are subscribed to the Google Groups "Machinekit" group. To unsubscribe from this group and stop receiving emails from it, send an email to machinekit+unsubscr...@googlegroups.com. Visit this group at https://groups.google.com/group/machinekit. For more options, visit https://groups.google.com/d/optout. -- website: http://www.machinekit.io blog: http://blog.machinekit.io github: https://github.com/machinekit --- You received this message because you are subscribed to the Google Groups "Machinekit" group. To unsubscribe from this group and stop receiving emails from it, send an email to machinekit+unsubscr...@googlegroups.com. Visit this group at https://groups.google.com/group/machinekit. For more options, visit https://groups.google.com/d/optout.