From owner-freebsd-current Sun Jun 24 10:50:19 2001 Delivered-To: freebsd-current@freebsd.org Received: from sax.sax.de (sax.sax.de [193.175.26.33]) by hub.freebsd.org (Postfix) with ESMTP id 5D55937B401 for ; Sun, 24 Jun 2001 10:50:08 -0700 (PDT) (envelope-from j@uriah.heep.sax.de) Received: (from uucp@localhost) by sax.sax.de (8.9.3/8.9.3) with UUCP id TAA12301 for freebsd-current@freebsd.org; Sun, 24 Jun 2001 19:50:05 +0200 (CEST) Received: (from j@localhost) by uriah.heep.sax.de (8.11.4/8.11.4) id f5OHUHj27111; Sun, 24 Jun 2001 19:30:17 +0200 (MET DST) (envelope-from j) Date: Sun, 24 Jun 2001 19:30:17 +0200 (MET DST) Message-Id: <200106241730.f5OHUHj27111@uriah.heep.sax.de> Mime-Version: 1.0 X-Newsreader: knews 1.0b.1 Reply-To: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch) Organization: Private BSD site, Dresden X-Phone: +49-351-2012 669 X-PGP-Fingerprint: DC 47 E6 E4 FF A6 E9 8F 93 21 E0 7D F9 12 D6 4E References: <200106211632.f5LGWWV90008@harmony.village.org> <20010621214747.061293808@overcee.netplex.com.au> From: j@uriah.heep.sax.de (Joerg Wunsch) Subject: Re: hints code bugs [Was: orm driver itches...] X-Original-Newsgroups: local.freebsd.current To: freebsd-current@freebsd.org Content-Transfer-Encoding: 8bit Content-Type: multipart/mixed; boundary="=-=-=__d4A40WGTWhScgOUopldlWVGGa__=-=-=" Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG --=-=-=__d4A40WGTWhScgOUopldlWVGGa__=-=-= Peter Wemm wrote: > Definately a bug. :-( The scanner is effectively concatenating > the device.hints and the config hints together when enumerating them. Speaking about bugs in the hints processing: Triggered by Michael Reifenberger , i started experimenting to get fdc(4) working as a KLD. There are several resource deallocation things still wrong in it, but i have fixed a number of them in my private tree (and know about more of them...). So far, i've encountered two serious problems. First, when trying to load the KLD at run-time, it cannot allocate the IO ports for fdc0, and then registers successfully as fdc1 (but of course without any drive connected to it since there are none mentioned in the hints for fdc1). Michael wrote that loading the KLD by the boot loader works well, however. I gdb'ed a little around, and found that the resource allocator will eventually divert int pci_alloc_resources() when called on behalf of fdc0. Naturally, this function doesn't even know how to handle IO port allocation, thus fails. When called next time on behalf of fdc1, the device is however (mis?)detected as PnP, and all works out well. Does anybody have an idea why the code would believe fdc0 to be connected to the PCI bus? (The hints do say it's located on the ISA bus, of course.) Can anybody describe the rough picture how the `default' IO port allocation (range 0 .. ~0UL) is supposed to work at all? I'm a bit confused. Is it possible to modify the hints at run-time (i. e. before kldloading the module)? Second, the driver doesn't deallocate completely. When calling devinfo(1) after kldunloading it, i get a panic somewhere in the depths of kprintf(), where it is trying to print the device description of fdc1 (where the string dev->descr was pointing to has been kldunloaded since). What needs to be done in order to destroy the device structure completely, and which of those tasks is the kernel linker supposed to perform? In case anybody wants to give it a try, i'll append my private patches collected so far. -- cheers, J"org .-.-. --... ...-- -.. . DL8DTL http://www.sax.de/~joerg/ NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-) --=-=-=__d4A40WGTWhScgOUopldlWVGGa__=-=-= Content-Type: text/plain Content-Disposition: attachment; filename="fdcdiff" Index: isa/fd.c =================================================================== RCS file: /home/ncvs/src/sys/isa/fd.c,v retrieving revision 1.204 diff -u -u -r1.204 fd.c --- isa/fd.c 2001/06/20 20:21:55 1.204 +++ isa/fd.c 2001/06/24 17:28:58 @@ -170,6 +170,8 @@ struct callout_handle toffhandle; struct callout_handle tohandle; struct devstat device_stats; + eventhandler_tag clonetag; + dev_t masterdev; device_t dev; fdu_t fdu; }; @@ -497,7 +499,7 @@ fdc_alloc_resources(struct fdc_data *fdc) { device_t dev; - int ispnp, ispcmcia; + int ispnp, ispcmcia, nports; dev = fdc->fdc_dev; ispnp = (fdc->flags & FDC_ISPNP) != 0; @@ -516,12 +518,13 @@ * uses the register with offset 6 for pseudo-DMA, and the * one with offset 7 as control register. */ + nports = ispcmcia ? 8 : (ispnp ? 1 : 6); fdc->res_ioport = bus_alloc_resource(dev, SYS_RES_IOPORT, &fdc->rid_ioport, 0ul, ~0ul, - ispcmcia ? 8 : (ispnp ? 1 : 6), - RF_ACTIVE); + nports, RF_ACTIVE); if (fdc->res_ioport == 0) { - device_printf(dev, "cannot reserve I/O port range\n"); + device_printf(dev, "cannot reserve I/O port range (%d ports)\n", + nports); return ENXIO; } fdc->portt = rman_get_bustag(fdc->res_ioport); @@ -569,7 +572,7 @@ 0ul, ~0ul, 1, RF_ACTIVE); if (fdc->res_ctl == 0) { device_printf(dev, - "cannot reserve control I/O port range\n"); + "cannot reserve control I/O port range (control port)\n"); return ENXIO; } fdc->ctlt = rman_get_bustag(fdc->res_ctl); @@ -760,8 +763,10 @@ return (error); } +#endif /* NCARD > 0 */ + static int -fdc_pccard_detach(device_t dev) +fdc_detach(device_t dev) { struct fdc_data *fdc; int error; @@ -772,6 +777,12 @@ if ((error = bus_generic_detach(dev))) return (error); + /* reset controller, turn motor off */ + fdout_wr(fdc, 0); + + if ((fdc->flags & FDC_NODMA) == 0) + isa_dma_release(fdc->dmachan); + if ((fdc->flags & FDC_ATTACHED) == 0) { device_printf(dev, "already unloaded\n"); return (0); @@ -785,8 +796,6 @@ return (0); } -#endif /* NCARD > 0 */ - /* * Add a child device to the fdc controller. It will then be probed etc. */ @@ -877,7 +886,7 @@ /* Device interface */ DEVMETHOD(device_probe, fdc_probe), DEVMETHOD(device_attach, fdc_attach), - DEVMETHOD(device_detach, bus_generic_detach), + DEVMETHOD(device_detach, fdc_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(device_suspend, bus_generic_suspend), DEVMETHOD(device_resume, bus_generic_resume), @@ -904,7 +913,7 @@ /* Device interface */ DEVMETHOD(device_probe, fdc_pccard_probe), DEVMETHOD(device_attach, fdc_attach), - DEVMETHOD(device_detach, fdc_pccard_detach), + DEVMETHOD(device_detach, fdc_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), DEVMETHOD(device_suspend, bus_generic_suspend), DEVMETHOD(device_resume, bus_generic_resume), @@ -1138,9 +1147,9 @@ cdevsw_add(&fd_cdevsw); /* XXX */ cdevsw_add_done++; } - EVENTHANDLER_REGISTER(dev_clone, fd_clone, 0, 1000); - make_dev(&fd_cdevsw, (fd->fdu << 6), - UID_ROOT, GID_OPERATOR, 0640, "fd%d", fd->fdu); + fd->clonetag = EVENTHANDLER_REGISTER(dev_clone, fd_clone, 0, 1000); + fd->masterdev = make_dev(&fd_cdevsw, (fd->fdu << 6), + UID_ROOT, GID_OPERATOR, 0640, "fd%d", fd->fdu); /* * Export the drive to the devstat interface. @@ -1158,6 +1167,10 @@ struct fd_data *fd; fd = device_get_softc(dev); + devstat_remove_entry(&fd->device_stats); + destroy_dev(fd->masterdev); + /* XXX need to destroy cloned devs as well */ + EVENTHANDLER_DEREGISTER(dev_clone, fd->clonetag); untimeout(fd_turnoff, fd, fd->toffhandle); return (0); Index: modules/Makefile =================================================================== RCS file: /home/ncvs/src/sys/modules/Makefile,v retrieving revision 1.186 diff -u -u -r1.186 Makefile --- modules/Makefile 2001/06/14 15:15:54 1.186 +++ modules/Makefile 2001/06/19 21:47:36 @@ -7,7 +7,7 @@ .endif SUBDIR= 3dfx accf_data accf_http agp aha amr an aue \ - cam ccd cd9660 coda cue dc de dgm digi ed fdescfs fs fxp if_disc if_ef \ + cam ccd cd9660 coda cue dc de dgm digi ed fdc fdescfs fs fxp if_disc if_ef \ if_ppp if_sl if_tap if_tun ip6fw ipfilter ipfw ispfw joy kue lge \ libmchain linux lnc md mii mlx msdosfs ncp netgraph nfs nge ntfs \ nullfs nwfs pcn portalfs procfs ${_random} \ --- /dev/null Sun Jun 24 19:19:47 2001 +++ modules/fdc/Makefile Tue Jun 19 23:44:45 2001 @@ -0,0 +1,28 @@ +# $FreeBSD$ + +.PATH: ${.CURDIR}/../../isa + +KMOD= fdc +SRCS= fd.c \ + opt_fdc.h card.h \ + bus_if.h card_if.h device_if.h isa_if.h +NOMAN= + +FDC_DEBUG= 1 # 0/1 +FDC_PCCARD= 0 # 0/1 whether pccard support (i. e. Y-E DATA PCMCIA fdc) is desired + +CLEANFILES= card.h + +opt_fdc.h: + touch ${.TARGET} +.if ${FDC_DEBUG} > 0 + echo "#define FDC_DEBUG 1" >> ${.TARGET} +.endif + +card.h: + touch ${.TARGET} +.if ${FDC_PCCARD} > 0 + echo "#define NCARD 1" >> ${.TARGET} +.endif + +.include --=-=-=__d4A40WGTWhScgOUopldlWVGGa__=-=-=-- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message