Am 27.08.2013 00:40, schrieb Paolo Bonzini: > Il 26/08/2013 19:15, Andreas Färber ha scritto: >>>> PCI devices are generally configurable, so you need to add prompts to them. >> IndustryPack is really misplaced in hw/char/ and I believe I posted >> patches to remedy that and let one actually find it in our source tree. >> There were no objections against hw/ipack/, alternatively it could go >> into hw/gpio/. (Currently my patch series is waiting to be respun due to >> changed QOM realize requirements from Anthony.) >> >> That having being said, IndustryPack does not depend on PCI, only the >> TPCI2000(?) PCI-IndustryPack bridge does. > > Both of them are under the same symbol right now. After all any of the > two is basically unusable without the other, and plans for extension > seem not to exist as even Linux has only that one bridge and one device. > > I have no objection to hw/ipack, but I have a question. Would you > follow the SCSI/USB model (with devices under hw/ipack, also followed > for IndustryPack in the Linux kernel) or the virtio model (where the > device remains under hw/char)? Generally we've tried to follow Linux > for hw/ structure unless maintainers preferred otherwise, so it would > prefer the former.
My quest is a) consistency and b) easily finding QOM base device classes for refactorings. PCI and USB were done before your big hw/ reorganization, and the biggest part of devices appears to follow the categorization by function (which is why I saw the overlap with Marcel's category markup). ipoctal232 looks correct in hw/char/ to me, so that it can benefit from any general char device refactorings. Andreas -- SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg
