Hello, Sitsofe. On Sat, Aug 13, 2016 at 09:56:17AM +0100, Sitsofe Wheeler wrote: > I was browsing through the kernel while looking at > https://github.com/systemd/systemd/issues/3943#issuecomment-239608897 > and I noticed that in libata-scsi.c shost->max_channel is set to 1. That number traces back to the original implementation when there was neither PMP support or the sysfs knobs. I don't know why it was set to 1 as 0 seems like the right number. Probably a mistake thinking that the number means the number of channels. > Further, if a SATA port multiplier is being used aren't channels used > to represent the port multiplier ports > (http://lxr.free-electrons.com/source/drivers/ata/libata-scsi.c?v=4.6#L3726 > )? If so does max_channel need to be increased to one less than > SATA_PMP_MAX_PORTS? If so, would this explain the results seen in Yeah, that sounds right. I don't have a working PM anymore and can't test it. It'd be great if you can spin up the patch and test it if you have one available. > https://www.usenix.org/system/files/fastpw13-paper7_0.pdf ? libata EH works mostly independent from SCSI layer and the channel thing could affect is only the sysfs representation part. As for the paper, a lot of port multipliers just show shitty error handling behavior. Back in the day, I tested quite a few of early hardware while developing for it and it was just a nightmare. PMs didn't really catch on and the PMs didn't have the chance to mature properly. Thanks. -- tejun -- To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html