[freebsd-amd64] Re: [freebsd-amd64] Possible 3ware 8506-12SATA Controller issues...

[Vinod Kashyap]

I can't speak for Victor, but I did talk to him this afternoon,
and he (and other support people) couldn't recall telling anyone
that 3ware supports FreeBSD amd64.  There might have been some miscommunication/misunderstandings since 3ware does support amd64
on Linux and Windows.  Even in the thread below, I don't see
him mention FreeBSD amd64.  Of course, the problem you had was
on amd64, and I don't know if he realized it.  The "different OS
versions" probably refers to 4.x and 5.x.

Anyways, I am sorry about any miscommunications, but as far as my
knowledge goes, there's not even been a FreeBSD amd64 version of the
driver built here, let alone tested.  As about future plans, it
might happen if there's sufficient interest, but at this point,
there's none.

Also, I wonder why you ran into build problems!  twa has been
in /sys/amd64/conf/GENERIC for sometime now, and no build problems
have been reported so far.


> -----Original Message-----
> From: mike [mailto:mike at mike2k.com]
> Sent: Tuesday, June 22, 2004 6:35 PM
> To: Peter Losher
> Cc: Vinod Kashyap; freebsd-amd64 at freebsd.org;
> freebsd-hardware at freebsd.org
> Subject: Re: [freebsd-amd64] Re: [freebsd-amd64] Possible 
> 3ware 8506-12
> SATA Controller issues...
> 
> 
> Here is the last response I received from Victor:
> 
> On Wed, 16 Jun 2004, Victor Pacheco wrote:
> 
> > Hi, Mike
> >
> >
> > I am told by our engineer 5.2.1 is not supported nor we support
> recompiling against newer kernels.  He wrote:
> > "FreeBSD is unlike Linux in that you can't use driver 
> source to compile
> a driver for different OS versions.
> Customers must have the compiled version of the driver for 
> their except
> FreeBSD version or it will not work.
> > Also, it was my understanding that the 9500S series does 
> not yet support
> FreeBSD.  The driver is not on the
> shipping CD or the web."
> >
> > I have passed this on to my manager (this does not make 
> sense) since we
> have the released driver/source.
> > I need to know if it should compile with 5.2.1.
> >
> > I just shot the email and wait to here back.
> >
> > Best Regards,
> >
> > Victor Pacheco
> > Technical Support Engineer
> 
> And here is the error I emailed him. I followed the 
> instructions from the
> manual, which were slightly outdated. Like it says 
> "twa_opt.h" instead of
> "opt_twa.h" - but I did everything I could to compile this far.
> 
> > [root at raid01 twa]# ls -1 /usr/src/sys/dev/twa
> > Makefile
> > twa.c
> > twa.h
> > twa_cam.c
> > twa_externs.h
> > twa_freebsd.c
> > twa_fwimg.c
> > twa_globals.c
> > twa_includes.h
> > twa_ioctl.h
> > twa_reg.h
> >
> > [root at raid01 twa]# grep twa /usr/src/sys/conf/options
> > TWA_DEBUG               opt_twa.h
> >
> > [root at raid01 twa]# grep twa /usr/src/sys/conf/files.amd64
> > dev/twa/twa.c           optional twa
> > dev/twa/twa_cam.c       optional twa
> > dev/twa/twa_freebsd.c   optional twa
> > dev/twa/twa_fwimg.c     optional twa
> > dev/twa/twa_globals.c   optional twa
> >
> > [root at raid01 twa]# grep twa /usr/src/sys/amd64/conf/RAID01
> > device          twa             # 3ware ATA RAID
> >
> > i get an error trying to compile it. if i keep the 
> TWA_DEBUG out of the
> > options file, when i try to "make depend" it says 
> "opt_twa.h not found"
> -
> > even though i don't want debugging. so i have to put that in and it
> shuts
> > up.
> >
> > here's the error:
> >
> > [root at raid01 conf]# config RAID01
> > Kernel build directory is ../compile/RAID01
> > Don't forget to do a ``make depend''
> > [root at raid01 conf]# cd ../compile/RAID01
> > [root at raid01 RAID01]# make depend
> > rm -f .olddep
> > if [ -f .depend ]; then mv .depend .olddep; fi
> > make _kernel-depend
> > if [ -f .olddep ]; then mv .olddep .depend; fi
> > rm -f .newdep
> > make -V CFILES -V SYSTEM_CFILES -V GEN_CFILES -V GEN_M_CFILES |
> > MKDEP_CPP="cc -E" CC="cc" xargs mkdep -a -f .newdep -O -pipe  -Wall
> > -Wredundant-decls -Wnested-externs -Wstrict-prototypes
> > -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual
> > -fformat-extensions -std=c99  -nostdinc -I-  -I. -I../../..
> > -I../../../contrib/dev/acpica -I../../../contrib/ipfilter
> > -I../../../contrib/dev/ath -I../../../contrib/dev/ath/freebsd
> > -I../../../contrib/ngatm -D_KERNEL -include opt_global.h -fno-common
> > -finline-limit=20000 -fno-strict-aliasing  -mcmodel=kernel 
> -mno-red-zone
> > -mfpmath=387 -mno-sse -mno-sse2 -mno-mmx -mno-3dnow  -msoft-float
> > -fno-asynchronous-unwind-tables -ffreestanding
> > make -V SFILES |  MKDEP_CPP="cc -E" xargs mkdep -a -f .newdep -x
> > assembler-with-cpp -DLOCORE -O -pipe  -Wall -Wredundant-decls
> > -Wnested-externs -Wstrict-prototypes  -Wmissing-prototypes
> -Wpointer-arith
> > -Winline -Wcast-qual  -fformat-extensions -std=c99  
> -nostdinc -I-  -I.
> > -I../../.. -I../../../contrib/dev/acpica -I../../../contrib/ipfilter
> > -I../../../contrib/dev/ath -I../../../contrib/dev/ath/freebsd
> > -I../../../contrib/ngatm -D_KERNEL -include opt_global.h -fno-common
> > -finline-limit=20000 -fno-strict-aliasing  -mcmodel=kernel 
> -mno-red-zone
> > -mfpmath=387 -mno-sse -mno-sse2 -mno-mmx -mno-3dnow  -msoft-float
> > -fno-asynchronous-unwind-tables -ffreestanding
> > rm -f .depend
> > mv .newdep .depend
> > [root at raid01 RAID01]# make
> > linking kernel
> > twa_cam.o: In function `twa_cam_setup':
> > twa_cam.o(.text+0x12): undefined reference to `cam_simq_alloc'
> > twa_cam.o(.text+0x58): undefined reference to `cam_sim_alloc'
> > twa_cam.o(.text+0x70): undefined reference to `cam_simq_free'
> > twa_cam.o(.text+0x8d): undefined reference to `xpt_bus_register'
> > twa_cam.o(.text+0xa2): undefined reference to `cam_sim_free'
> > twa_cam.o(.text+0xd4): undefined reference to `xpt_create_path'
> > twa_cam.o(.text+0xe8): undefined reference to `xpt_bus_deregister'
> > twa_cam.o(.text+0xf9): undefined reference to `cam_sim_free'
> > twa_cam.o(.text+0x114): undefined reference to `xpt_setup_ccb'
> > twa_cam.o(.text+0x143): undefined reference to `xpt_action'
> > twa_cam.o: In function `twa_cam_detach':
> > twa_cam.o(.text+0x176): undefined reference to `xpt_free_path'
> > twa_cam.o(.text+0x191): undefined reference to `xpt_bus_deregister'
> > twa_cam.o(.text+0x1a2): undefined reference to `cam_sim_free'
> > twa_cam.o: In function `twa_execute_scsi':
> > twa_cam.o(.text+0x408): undefined reference to `xpt_done'
> > twa_cam.o: In function `twa_action':
> > twa_cam.o(.text+0x58f): undefined reference to `cam_calc_geometry'
> > twa_cam.o(.text+0x650): undefined reference to `xpt_done'
> > twa_cam.o: In function `twa_request_bus_scan':
> > twa_cam.o(.text+0x6b9): undefined reference to `xpt_periph'
> > twa_cam.o(.text+0x6ca): undefined reference to `xpt_create_path'
> > twa_cam.o(.text+0x6e0): undefined reference to `xpt_setup_ccb'
> > twa_cam.o(.text+0x701): undefined reference to `xpt_action'
> > twa_cam.o: In function `twa_bus_scan_cb':
> > twa_cam.o(.text+0x730): undefined reference to `xpt_free_path'
> > twa_cam.o: In function `twa_scsi_complete':
> > twa_cam.o(.text+0x87b): undefined reference to `xpt_done'
> > twa_cam.o: In function `twa_drain_busy_queue':
> > twa_cam.o(.text+0x8d9): undefined reference to `xpt_done'
> > twa_cam.o: In function `twa_disallow_new_requests':
> > twa_cam.o(.text+0x951): undefined reference to `xpt_freeze_simq'
> > *** Error code 1
> >
> > Stop in /usr/src/sys/amd64/compile/RAID01.
> 
> 
> 
> 
> 
> On Tue, 22 Jun 2004, Peter Losher wrote:
> 
> >
> > Any word on work on a amd64-compatible driver? (mike seemed 
> to indicate that a
> > amd64 driver for the 9xxx series might come later this summer?)
> >
> >
> > Tried compiling a 5.2.1 kernel w/ the twe driver plopped 
> from -CURRENT
> > cvsup'ed 30 minutes ago, and it didn't go to well:
> >
> > -=-
> >
> > If there is something I can hack in to twe_freebsd.c to get 
> it to compile
> > under 5.2.1, I am all ears.
> >
> 
> I myself would love this. I'd pay $250 cash to expedite this 
> fix. Someone
> could work directly with someone at 3ware and get 9500 matured, at
> least... I wonder if I tried -current if my 9500 would work 
> better. Is it
> still based off the 7xxx series driver though?
> 
> At this point I've got hardware and software can't handle it. 
> It's sort of
> discouraging, usually it's the opposite. I might pay up to $500 to get
> this going. It's worth it that much to get this damn thing 
> fixed so I can
> send this box back east and get it into production =)
> 
> 
> 
> > -Peter
> > --
> > Peter_Losher at isc.org | ISC | OpenPGP 0xE8048D08 | "The bits 
> must flow"
> >
>