From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tom Denchfield Subject: Re: CF Card Adapter White List Candidate Date: Wed, 5 Jan 2011 21:33:53 -0800 (PST) Message-ID: <186863.99156.qm@web112005.mail.gq1.yahoo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from nm8.bullet.mail.ne1.yahoo.com ([98.138.90.71]:29549 "HELO nm8.bullet.mail.ne1.yahoo.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1752486Ab1AFFdz convert rfc822-to-8bit (ORCPT ); Thu, 6 Jan 2011 00:33:55 -0500 Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Robert Hancock Cc: Tejun Heo , linux-ide@vger.kernel.org Hello, Robert, it looks like I may have found an answer to my UDMA speed quest= ion (see below) that I asked earlier today.=A0 URL03 below has a pictur= e (top-right) of an Addonics Dual CF-IDE HDD Adapter that looks to me l= ike it can be connected to a motherboard IDE socket via a 80-wire IDE c= able.=A0 (This adapter's IDE socket is not shown in the picture. On se= cond thought, maybe this device does not have a 40-pin IDE socket since= the article is about laptop computers.)=A0 The article says to use a (= now) outdated patch, even with, I would guess, an 80-conductor IDE cabl= e, to 'force 80c.'=A0 With the outdated patch, URL03 has a dmesg excerp= t in a table that shows a UDMA/100 CF card being configured by Linux as= a UDMA/100 device. =20 Title: Compact Flash boot drive URL03: http://www.thinkwiki.org/wiki/Compact_Flash_boot_drive With my plug-in IDE adapter (pictured in URL01 below) that is plugged d= irectly into a desktop computer's motherboard IDE socket, my UDMA/66 CF= Card is configured by the Linux version 2.6.35-22-generic kernel that = is used by the Ubuntu v10.10 LiveCD that I use, as a UDMA/33 device.=A0= However, if I add libata.force=3D80c as a boot parameter, as recommend= ed by Tejun, Linux configures my CF card as a UDMA/66 device. (In the first email that I sent to linux-ide@vger.kernel.org, I attache= d a dmesg demonstrating that the added boot parameter produced the UDMA= /66 configuration.=A0 I attached two text files that I think totaled ab= out 65KB, and maybe linux-ide@vger.kernel.org strips files of this size= from emails.=A0 I also do not expect anyone to remember that far back = when answering my questions.) I do not know whether the following dmesg is the one that I originally = sent, but it is typical of what happens with the liabat.forec=3D80c par= ameter: [=A0 =A0 0.000000] Kernel command line: file=3D/cdrom/preseed/ubuntu.se= ed boot=3Dcasper initrd=3D/casper/initrd.lz radeon.modeset=3D0 libata.f= orce=3D80c toram -- =2E [=A0 =A0 5.088368] ata3: FORCE: cable set to 80c [=A0 =A0 5.088452] ata3.00: CFA: TS8GCF133, 20100804, max UDMA/66 [=A0 =A0 5.088509] ata3.00: 15662304 sectors, multi 0: LBA=20 [=A0 =A0 5.104301] ata3.00: configured for UDMA/66 =2E [=A0 =A0 5.564449] scsi 2:0:0:0: Direct-Access=A0 =A0=A0=A0ATA=A0 =A0 =A0= TS8GCF133=A0 =A0 =A0 =A0 2010 PQ: 0 ANSI: 5 [=A0 =A0 5.564819] sd 2:0:0:0: Attached scsi generic sg1 type 0 [=A0 =A0 5.565631] sd 2:0:0:0: [sdb] 15662304 512-byte logical blocks: = (8.01 GB/7.46 GiB) The reason that I mentioned powering my CF adapter with a floppy drive = power plug is that the first time I booted after inserting the adapter,= I forgot to plug the floppy drive power plug into the adapter.=A0 The = adapter's power on LED did not light, and I think that Ubuntu did not s= ee the CF card. The reason that I mentioned the pin that was intentionally left out of = the motherboard IDE socket was that I thought that the missing pin migh= t have had something to do with needing to power the adapter with the f= loppy drive power plug given my experience with forgetting to plug the = floppy power plug into the adapter as mentioned in the previous paragra= ph. The pin that is missing from the motherboard IDE socket is very close t= o the middle of the IDE socket.=A0 My guess is that the pin is in the i= n the range #18 to #24.=A0 It is far away from pin #34.=A0 I am going b= y memory here, but I clearly remember that the missing pin is in the mi= ddle.=A0 Pin #34 is going into the CF card adapter when I insert it.=A0= I do not know whether pin #34 is grounded at the mainboard. When I asked my question, before I noticed and paid any real attention = to the UDMA/100 CF card configured as UDMA/100 in URL03 above, I was co= ncerned that there might be a UDMA/66 speed limit with the type of CF c= ard adapter that I have. We may never know for sure until someone who = has a CF card that is faster than UDMA/66 tries it in the URL01 adapter= with Linux and the libata.force=3D80c boot parameter. Robert, thank you for helping me. I hope that my additional comments a= nd the information above will contribute to further Linux kernel develo= pment. =A0=20 Tom --- On Wed, 1/5/11, Robert Hancock wrote: > From: Robert Hancock > Subject: Re: CF Card Adapter White List Candidate > To: "Tom Denchfield" > Cc: "Tejun Heo" , linux-ide@vger.kernel.org > Date: Wednesday, January 5, 2011, 5:47 PM > On Wed, Jan 5, 2011 at 12:18 PM, Tom > Denchfield > wrote: > > Hello, > > > > Tejun, this CF card adapter plugs directly into an IDE > socket (Channel 0, in my case) on a desktop computer > motherboard.=A0 You can see a picture of the adapter in the > following URL01: > > > > URL01: http://preview.tinyurl.com/233u92u > > > > I have to power the adapter with a floppy drive power > plug because my motherboard's IDE sockets were intentionally > manufactured without the pin that could have supplied power > through an IDE cable. > > > > Does anyone have a top of the head opinion as to > whether there are inherent usage limitations in either this > passive CF card adapter and/or in the Linux kernel used with > Ubuntu v10.10 that would prevent users with CF cards faster > than UDMA/66 (see URL02 below) from being configured by > Linux to work at speeds above UDMA/66? =A0The reason that I > ask this question is that I bought both my UDMA/66 CF card > and the adapter from the URL01 seller.=A0 It is foreseeable > that a lot of customers might have purchased faster CF cards > envisioning that they would work at speeds higher than > UDMA/66 with the adapter.=A0 The 'Technical Details' section > of URL01 does not state anything about UDMA speed usage > limitations. > > > > URL02: http://www.hjreggel.net/cardspeed/index.html > > > > I think that there are CF card adapters on eBay that > are similar to the one pictured in URL01 above, and there > are also IDE adapters that will hold two CF cards. >=20 > The reason the speed is being limited to 33 is that the > cable > detection code is detecting that an 80-wire cable doesn't > appear to be > connected. In order for an 80-wire cable to be detected, > pin 34 on the > cable is supposed to be grounded at the motherboard end. In > the case > of this adapter, since it plugs directly into the > motherboard, it > would have to be grounding pin 34 to get an 80-wire > detection, but > presumably it's not. >=20 > > > > Tom > > > > --- On Wed, 1/5/11, Tejun Heo > wrote: > > > >> From: Tejun Heo > >> Subject: Re: CF Card Adapter White List Candidate > >> To: "Robert Hancock" > >> Cc: "Tom Denchfield" , > linux-ide@vger.kernel.org > >> Date: Wednesday, January 5, 2011, 5:51 AM > >> Hello, > >> > >> On Tue, Jan 04, 2011 at 10:19:17PM -0600, Robert > Hancock > >> wrote: > >> > On 01/04/2011 04:18 PM, Tom Denchfield > wrote: > >> >> As you further recommended, in an attempt > to get > >> the SD-CF-IDE-DI > >> >> IDE to CF Adapter that holds my CF card > white > >> listed, I am > >> >> submitting the attached > >> >> > >> > libata_force_80c_CF_card_adapter_whitelist_candidate_information.txt > >> >> file that has the output from executing > four > >> terminal commands to > >> >> hopefully supply enough information to > get it > >> whitelisted. > >> > >> Thanks. > >> > >> >> Although I expect that my RV280 Radeon > 9200 Pro > >> video card is too > >> >> old to expend much effort on, it would be > nice to > >> also get it > >> >> whitelisted so that future newbie Linux > users who > >> do not know about > >> >> using the radeon.modeset=3D0 parameter on > the kernel > >> command line > >> >> will not be looking at a black screen > with a > >> blinking cursor when > >> >> they attempt to use a LiveCD for > troubleshooting, > >> or whatever. > >> >> > >> >> I purposely did not include the URL of > one of the > >> Internet sites > >> >> that sells the adapter in this email in > case this > >> is a no no. > >> > >> I don't think that's a no no unless the intention > is > >> commercial. > >> > >> >> Tejun, I am not sure that I will have > either the > >> time or the > >> >> interest to learn the intricacies of > editing > >> >> http://www.thinkwiki.org/wiki/Compact_Flash_boot_drive > >> to add > >> >> libata.force=3D80c to this site in lieu of > >> force_cbl=3D80: so that > >> >> fewer people will be asking you > questions.=A0 I > >> have a ThinkPad > >> >> without a HDD plus a laptop adapter that > will hold > >> my CF card, but > >> >> my Think Pad is a lot older than the > ones > >> discussed on > >> >> Compact_Flash_boot_drive.=A0 In addition, > I did > >> not see any recent > >> >> updates to this Internet page, but maybe > I can > >> find someone to > >> >> email who will update this page. > >> > >> Yeah, just ping someone. > >> > >> > Is there actually any way to identify the > adapter > >> automatically? > >> > AFAIK, these CF-IDE adapters are just > passive > >> circuitry and there's > >> > no way to identify them through software. > >> > >> Hmmm... I was thinking this was somehing > integrated to the > >> machine (so > >> the dmidecode), in which case we can combine dmi + > pci > >> function # + > >> port # to whitelist the device, which we already > do for > >> some type of > >> quirks.=A0 If it's a plug in device, we obviously > can't > >> identify it. > >> Tom, is the CF slot integrated to the machine or > is it > >> something you > >> put into the PCMCIA slot? > >> > >> Thanks. > >> > >> -- > >> tejun > >> > > > > > > > > > =20