delkin_cb: new driver for Cardbus IDE CF adaptor
PATCH 1/1 delkin_cb: new driver for Cardbus IDE CF adaptor
This is the first release of a PCI-IDE module to support
the Delkin/ASKA/Workbit Cardbus CompactFlash<-->IDE adaptor. It may (or not) also work for other versions of this hardware (eg. SD).
This module is based on the existing ide-cs module, and works fine for me.
Patch is against 2.4.28-pre5-bk6, and it would be good if possible to include this in the next issue of 2.4.xx (Marcelo?).
An equivalent patch for 2.6.xx is being worked on.
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
drivers/ide/Config.in +1 -0
@@ -18,6 +18,7 @@ dep_mbool ' Auto-Geometry Resizing support' CONFIG_IDEDISK_STROKE $CONFIG_BLK_DEV_IDEDISK dep_tristate ' PCMCIA IDE support' CONFIG_BLK_DEV_IDECS $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA + dep_tristate ' Cardbus IDE support (Delkin/ASKA/Workbit)' CONFIG_BLK_DEV_DELKIN $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA $CONFIG_PCI dep_tristate ' Include IDE/ATAPI CDROM support' CONFIG_BLK_DEV_IDECD $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI TAPE support' CONFIG_BLK_DEV_IDETAPE $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI FLOPPY support' CONFIG_BLK_DEV_IDEFLOPPY $CONFIG_BLK_DEV_IDE
drivers/ide/pci/delkin_cb.c +147 -0
147 lines not on this page · raw message
drivers/ide/pci/Makefile +1 -0
@@ -34,6 +34,7 @@ obj-$(CONFIG_BLK_DEV_TRM290) += trm290.o obj-$(CONFIG_BLK_DEV_VIA82CXXX) += via82cxxx.o obj-$(CONFIG_BLK_DEV_TRIFLEX) += triflex.o +obj-$(CONFIG_BLK_DEV_DELKIN) += delkin_cb.o # Must appear at the end of the block obj-$(CONFIG_BLK_DEV_GENERIC) += generic.o
drivers/scsi/nsp32.h +0 -1
@@ -22,7 +22,6 @@* VENDOR/DEVICE ID */ #define PCI_VENDOR_ID_IODATA 0x10fc -#define PCI_VENDOR_ID_WORKBIT 0x1145#define PCI_DEVICE_ID_NINJASCSI_32BI_CBSC_II 0x0005#define PCI_DEVICE_ID_NINJASCSI_32BI_KME 0xf007
include/linux/pci_ids.h +3 -0
@@ -2057,3 +2057,6 @@#define PCI_DEVICE_ID_MICROGATE_USC 0x0010#define PCI_DEVICE_ID_MICROGATE_SCC 0x0020#define PCI_DEVICE_ID_MICROGATE_SCA 0x0030 + +#define PCI_VENDOR_ID_WORKBIT 0x1145 +#define PCI_DEVICE_ID_WORKBIT_CB 0xf021
please also cc: [email protected]
Quoted 1 line
An equivalent patch for 2.6.xx is being worked on.
generally it should be like that: 2.6.x first, 2.4.x later
Quoted 6 lines
+ * This is slightly peculiar, in that it is a PCI driver, + * but is NOT an IDE PCI driver -- the IDE layer does not directly + * support hot insertion/removal of PCI interfaces, so this driver + * is unable to use the IDE PCI interfaces. Instead, it uses the + * same interfaces as the ide-cs (PCMCIA) driver uses. + * On the plus side, the driver is also smaller/simpler this way.
IDE layer doesn't support hot insertion/removal of _any_ interfaces
ide_unregister() calls are not allowed unless somebody fixes locking (Alan fixed many issues but some more work is still needed)
Bartlomiej
Okay, patch withdrawn.
I'll just apply it to my own kernels for my own use.
Whatever happended to the days when Linux *wanted* more drivers and such?
Oh well..
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Quoted 28 lines · Bartlomiej Zolnierkiewicz
please also cc: [email protected]
Quoted 1 line
An equivalent patch for 2.6.xx is being worked on.
generally it should be like that: 2.6.x first, 2.4.x later
Quoted 6 lines
+ * This is slightly peculiar, in that it is a PCI driver, + * but is NOT an IDE PCI driver -- the IDE layer does not directly + * support hot insertion/removal of PCI interfaces, so this driver + * is unable to use the IDE PCI interfaces. Instead, it uses the + * same interfaces as the ide-cs (PCMCIA) driver uses. + * On the plus side, the driver is also smaller/simpler this way.IDE layer doesn't support hot insertion/removal of _any_ interfaces
ide_unregister()calls are not allowed unless somebody fixes locking (Alan fixed many issues but some more work is still needed)Bartlomiej
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Hi Bartlomiej!
Quoted 2 lines · Bartlomiej Zolnierkiewicz
Just port it to 2.6.x...
From my current negative experiences with 2.6.xx, I'll pass. That kernel breaks suspend/resume on my notebooks, and is a dog for disk performance. Still, I'd happily port this new driver to it if there was a hope in hell that the effort wouldn't be a total waste of my time.
Quoted 1 line
ide_unregister()is disallowed, unless IDE locking is fixed
That just happens to be the existing interface used by the existing PCMCIA layer. The new delkin_cb driver simply does exactly the same calls to link in/out of the kernel as ide-cs does today.
I suppose that means we should remove ide-cs as well.
Cheers
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Quoted 3 lines · Mark Lord
Okay, patch withdrawn.
I'll just apply it to my own kernels for my own use.
Just port it to 2.6.x...
Quoted 2 lines
Whatever happended to the days when Linux *wanted* more drivers and such?
New drivers are still welcomed... but days of applying new drivers without any complaints are long gone... ;-)
Now speaking seriously:
* 2.4.x is deprecated (sorry, Marcelo ;)
* this driver shouldn't require much work to port it to 2.6.x
* ide_unregister() is disallowed, unless IDE locking is fixed
Cheers,
Bartlomiej
Quoted 2 lines · Mark Lord
Whatever happended to the days when Linux *wanted* more drivers and such?
I think Marcelo is trying to move 2.4 into maintenance mode. At least that's what I've seen around a few places.
Chris
Quoted 9 lines · Mark Lord
Hi Bartlomiej!
Quoted 2 lines · Bartlomiej Zolnierkiewicz
Just port it to 2.6.x...
From my current negative experiences with 2.6.xx, I'll pass.
That kernel breaks suspend/resume on my notebooks, and is a dog for disk performance. Still, I'd happily
bug reports / patches are the way to go
Quoted 2 lines
port this new driver to it if there was a hope in hell that the effort wouldn't be a total waste of my time.
2.4.x is a total waste of my time 8)
Quoted 5 lines
Quoted 1 line
ide_unregister()is disallowed, unless IDE locking is fixedThat just happens to be the existing interface used by the existing PCMCIA layer. The new
delkin_cbdriver simply does exactly the same calls to link in/out of the kernel as ide-cs does today.
"other drivers use it so it must be OK", no it is not it was discussed few times already...
Quoted 1 line
I suppose that means we should remove ide-cs as well.
Not adding new users is sufficient for now.
Cheers,
Bartlomiej
Quoted 3 lines · Mark Lord
Quoted 1 line
Not adding new users is sufficient for now.
It's amazing just how far we've come in ten years.
Could you explain?
Quoted 1 line
Not adding new users is sufficient for now.
It's amazing just how far we've come in ten years.
Cheers
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Quoted 4 lines · Bartlomiej Zolnierkiewicz
ide_unregister()calls are not allowed unless somebody fixes locking (Alan fixed many issues but some more work is still needed)From review I think I've got them all fixed in 2.6, but yes 2.4 is a
bit of a lost cause there (although ide-cs works generally so its no big barrier)
That makes a lot of sense :)
Quoted 19 lines · Alan Cox
Quoted 4 lines · Mark Lord
That kernel breaks suspend/resume on my notebooks, and is a dog for disk performance. Still, I'd happily port this new driver to it if there was a hope in hell that the effort wouldn't be a total waste of my time.
If you can drop it into the 2.6-ac patches that would be great and I will merge it from review of the current code. That should have correct IDE locking and also possibly correct PCI IDE locking (actually I need to fix one detail).
Having a cardbus IDE adapter would be a godsend for my testing so I'll see if I can find a source of them over here too for brutalising the code.
The only visible difference is that we pass a hwif to the unregister functions in the -ac tree.
Alan
Quoted 4 lines · Mark Lord
That kernel breaks suspend/resume on my notebooks, and is a dog for disk performance. Still, I'd happily port this new driver to it if there was a hope in hell that the effort wouldn't be a total waste of my time.
If you can drop it into the 2.6-ac patches that would be great and I will merge it from review of the current code. That should have correct IDE locking and also possibly correct PCI IDE locking (actually I need to fix one detail).
Having a cardbus IDE adapter would be a godsend for my testing so I'll see if I can find a source of them over here too for brutalising the code.
The only visible difference is that we pass a hwif to the unregister functions in the -ac tree.
Alan
Quoted 4 lines · Alan Cox
Quoted 2 lines
From review I think I've got them all fixed in 2.6, but yes 2.4 is a
bit of a lost cause there (although ide-cs works generally so its no big barrier)
That was my feeling on it too. Any races that remain in 2.4 are really a non-issue for removable PC-CARD devices -- the user is popping them out of the socket regardless, so it cannot really be any more harmful to tell the kernel they don't physically exist afterwards. :)
ide-cs does seem to be reliable, as is delkin_cb. It's not like these devices are proliferating -- USB flash readers are pretty much taking over on newer USB2 machines, and this unit is likely the last of the breed.
The adapter I have here is a "Delkin Devices Cardbus 32 eFilm PRO" compact flash adapter. They are also sold in Japan under the ASKA name, and the chipset is by Workbit. The only North American sources I know of for this card are B&H Photo in New York (they have a website) and Downtown Camera in Toronto.
These cards are popular among photographic professionals (and keen amateurs) because of the compact size, and under Windows they are much faster than many USB2 or Firewire readers. Under Linux, they just plain are not recognized -- unless delkin_cb is configured, in which case they currently achieve about 2/3 the speed of the MS driver.
I'll pull down your (Alan) latest tree and re-post a 2.6 patch against it. But I would really like to see Marcelo pick up the 2.4 version as well, since that is what people are using today.
Cheers
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Quoted 3 lines · Mark Lord
I'll pull down your (Alan) latest tree and re-post a 2.6 patch against it. But I would really like to see Marcelo pick up the 2.4 version as well, since that is what people are using today.
Therefore 2.4 version is OK with me.
Quoted 2 lines · Bartlomiej Zolnierkiewicz
Therefore 2.4 version is OK with me.
Great.
Marcelo, here is an updated 2.4 patch, which replaces the original. The changes are a spelling correction, and I've added a precautionary memset(0) on the hw struct before populating it (copied back from 2.6).
This is a driver for the Delkin/ASKA/Workbit cardbus CF adapter.
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Documentation/Configure.help +7 -0
@@ -781,6 +781,13 @@ <file:Documentation/modules.txt>. The module will be called ide-cs.o +Cardbus IDE support (Delkin/ASKA/Workbit) +CONFIG_BLK_DEV_DELKIN + Support for Delkin/ASKA/Workbit cardbus CompactFlash Adapters. + This may also work for similar SD and XD adapters. If you want + to be able to use one of these, then say M here. The module will + be called delkin_cb.o + Include IDE/ATAPI CD-ROM support CONFIG_BLK_DEV_IDECD If you have a CD-ROM drive using the ATAPI protocol, say Y. ATAPI is
drivers/ide/Config.in +1 -0
@@ -18,6 +18,7 @@ dep_mbool ' Auto-Geometry Resizing support' CONFIG_IDEDISK_STROKE $CONFIG_BLK_DEV_IDEDISK dep_tristate ' PCMCIA IDE support' CONFIG_BLK_DEV_IDECS $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA + dep_tristate ' Cardbus IDE support (Delkin/ASKA/Workbit)' CONFIG_BLK_DEV_DELKIN $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA $CONFIG_PCI dep_tristate ' Include IDE/ATAPI CDROM support' CONFIG_BLK_DEV_IDECD $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI TAPE support' CONFIG_BLK_DEV_IDETAPE $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI FLOPPY support' CONFIG_BLK_DEV_IDEFLOPPY $CONFIG_BLK_DEV_IDE
drivers/ide/pci/delkin_cb.c +149 -0
149 lines not on this page · raw message
drivers/ide/pci/Makefile +1 -0
@@ -34,6 +34,7 @@ obj-$(CONFIG_BLK_DEV_TRM290) += trm290.o obj-$(CONFIG_BLK_DEV_VIA82CXXX) += via82cxxx.o obj-$(CONFIG_BLK_DEV_TRIFLEX) += triflex.o +obj-$(CONFIG_BLK_DEV_DELKIN) += delkin_cb.o # Must appear at the end of the block obj-$(CONFIG_BLK_DEV_GENERIC) += generic.o
drivers/scsi/nsp32.h +0 -1
@@ -22,7 +22,6 @@* VENDOR/DEVICE ID */ #define PCI_VENDOR_ID_IODATA 0x10fc -#define PCI_VENDOR_ID_WORKBIT 0x1145#define PCI_DEVICE_ID_NINJASCSI_32BI_CBSC_II 0x0005#define PCI_DEVICE_ID_NINJASCSI_32BI_KME 0xf007
include/linux/pci_ids.h +3 -0
@@ -2057,3 +2057,6 @@#define PCI_DEVICE_ID_MICROGATE_USC 0x0010#define PCI_DEVICE_ID_MICROGATE_SCC 0x0020#define PCI_DEVICE_ID_MICROGATE_SCA 0x0030 + +#define PCI_VENDOR_ID_WORKBIT 0x1145 +#define PCI_DEVICE_ID_WORKBIT_CB 0xf021
wrt to 2.6.x version
drivers/ide/pci/delkin_cb.c line 81
Quoted 3 lines
drivers/ide/pci/delkin_cb.c +3
-0+memset (&hw,0 ,sizeof (hw)); +ide_init_hwif_ports (&hw, (ide_ioreg_t)(base +0x10 ), + (ide_ioreg_t)(base +0x1e ),NULL );
please use ide_std_init_ports()
drivers/ide/pci/delkin_cb.c line 87
Quoted 3 lines
drivers/ide/pci/delkin_cb.c +3
-0+ rc =ide_register_hw (&hw, &hwif); +if (rc <0 )/* ide_register_hw likes to be invoked twice (buggy) */ + rc =ide_register_hw (&hw, &hwif);
is this needed in 2.6.x and if so why?
Quoted 1 line
+
drive->id->csfo= 0; /* workaround foridedisk_openbug */
ditto
Quoted 1 line · Bartlomiej Zolnierkiewicz
wrt to 2.6.x version
..
Quoted 1 line
please use
ide_std_init_ports()
Okay, will do.
Quoted 5 lines
drivers/ide/pci/delkin_cb.c line 87
Quoted 3 lines
drivers/ide/pci/delkin_cb.c +3
-0+ rc =ide_register_hw (&hw, &hwif); +if (rc <0 )/* ide_register_hw likes to be invoked twice (buggy) */ + rc =ide_register_hw (&hw, &hwif);is this needed in 2.6.x and if so why?
Not sure yet -- still testing, though I've already done an #if 0 on it.
Quoted 1 line
Quoted 1 line
+
drive->id->csfo= 0; /* workaround foridedisk_openbug */
Not there in the 2.6.xx version.
And in 2.4.xx.. why is idedisk_open() examining vendor-specific fields of the IDENTIFY data, anyway? Very very unsafe.
I put the above one-liner workaround (drive->id->csfo) into delkin_cb to bypass the problems it creates for now, until idedisk_open gets fixed.
Normally I'd just send a patch to fix idedisk_open(), but since I don't even understand what it is trying to do, it would be safer for whoever put that code there to have a second look. Especially since 2.4.xx is supposed to be stable now -- if it ain't broke, don't break it. :)
Cheers
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Quoted 8 lines · Mark Lord
Quoted 1 line
Quoted 1 line
+
drive->id->csfo= 0; /* workaround foridedisk_openbug */Not there in the 2.6.xx version.
And in 2.4.xx.. why is
idedisk_open()examining vendor-specific fields of the IDENTIFY data, anyway? Very very unsafe.
I put the above one-liner workaround (drive->id->csfo) intodelkin_cbto bypass the problems it creates for now, untilidedisk_opengets fixed.
id->csfo is still checked in 2.6.x but in idedisk_setup()
Quoted 4 lines
Normally I'd just send a patch to fix
idedisk_open(), but since I don't even understand what it is trying to do, it would be safer for whoever put that code there to have a second look. Especially since 2.4.xx is supposed to be stable now -- if it ain't broke, don't break it. :)
I guess that Alan knows more about id->csfo.
Quoted 7 lines · Bartlomiej Zolnierkiewicz
Quoted 3 lines · Mark Lord
Okay, patch withdrawn.
I'll just apply it to my own kernels for my own use.
Just port it to 2.6.x...
Quoted 8 lines
Quoted 2 lines
Whatever happended to the days when Linux *wanted* more drivers and such?
New drivers are still welcomed... but days of applying new drivers without any complaints are long gone... ;-)
Now speaking seriously:
* 2.4.x is deprecated (sorry, Marcelo ;)
Buaaaahhh...
Hehe.
People still use 2.4 because they are not prepared (for whatever reason) to or do not want a v2.6 upgrade, but yes, I agree it is deprecated.
Quoted 1 line
* this driver shouldn't require much work to port it to 2.6.x
As I said before in my opinion drivers in new v2.4.x releases for new devices are OK as long as they meet a decent coding and quality standard and are well tested.
And, very importantly, benefit a substantial amount of users.
Having the driver in v2.6 first is pretty much required. It has the chance of being put under test, as well as passing Linus/Andrew "level of acceptance", plus reviewing from other people which is always good.
Mark, v2.4.28 is approaching -rc stage (-rc1 should be out tomorrow), we can include the driver during v2.4.29-pre.
In the meantime you can port it to v2.6 ?
Quoted 1 line
*
ide_unregister()is disallowed, unless IDE locking is fixed
Quoted 2 lines · Bartlomiej Zolnierkiewicz
Therefore 2.4 version is OK with me.
Great.
Marcelo, here is an updated 2.4 patch, which replaces the original. The changes are a spelling correction, and I've added a precautionary memset(0) on the hw struct before populating it (copied back from 2.6).
This is a driver for the Delkin/ASKA/Workbit cardbus CF adapter.
Signature · 2 lines
Mark Lord (hdparm keeper & the original "Linux IDE Guy")
Documentation/Configure.help +7 -0
@@ -781,6 +781,13 @@ <file:Documentation/modules.txt>. The module will be called ide-cs.o +Cardbus IDE support (Delkin/ASKA/Workbit) +CONFIG_BLK_DEV_DELKIN + Support for Delkin/ASKA/Workbit cardbus CompactFlash Adapters. + This may also work for similar SD and XD adapters. If you want + to be able to use one of these, then say M here. The module will + be called delkin_cb.o + Include IDE/ATAPI CD-ROM support CONFIG_BLK_DEV_IDECD If you have a CD-ROM drive using the ATAPI protocol, say Y. ATAPI is
drivers/ide/Config.in +1 -0
@@ -18,6 +18,7 @@ dep_mbool ' Auto-Geometry Resizing support' CONFIG_IDEDISK_STROKE $CONFIG_BLK_DEV_IDEDISK dep_tristate ' PCMCIA IDE support' CONFIG_BLK_DEV_IDECS $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA + dep_tristate ' Cardbus IDE support (Delkin/ASKA/Workbit)' CONFIG_BLK_DEV_DELKIN $CONFIG_BLK_DEV_IDE $CONFIG_PCMCIA $CONFIG_PCI dep_tristate ' Include IDE/ATAPI CDROM support' CONFIG_BLK_DEV_IDECD $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI TAPE support' CONFIG_BLK_DEV_IDETAPE $CONFIG_BLK_DEV_IDE dep_tristate ' Include IDE/ATAPI FLOPPY support' CONFIG_BLK_DEV_IDEFLOPPY $CONFIG_BLK_DEV_IDE
drivers/ide/pci/delkin_cb.c +149 -0
149 lines not on this page · raw message
drivers/ide/pci/Makefile +1 -0
@@ -34,6 +34,7 @@ obj-$(CONFIG_BLK_DEV_TRM290) += trm290.o obj-$(CONFIG_BLK_DEV_VIA82CXXX) += via82cxxx.o obj-$(CONFIG_BLK_DEV_TRIFLEX) += triflex.o +obj-$(CONFIG_BLK_DEV_DELKIN) += delkin_cb.o # Must appear at the end of the block obj-$(CONFIG_BLK_DEV_GENERIC) += generic.o
drivers/scsi/nsp32.h +0 -1
@@ -22,7 +22,6 @@* VENDOR/DEVICE ID */ #define PCI_VENDOR_ID_IODATA 0x10fc -#define PCI_VENDOR_ID_WORKBIT 0x1145#define PCI_DEVICE_ID_NINJASCSI_32BI_CBSC_II 0x0005#define PCI_DEVICE_ID_NINJASCSI_32BI_KME 0xf007
include/linux/pci_ids.h +3 -0
@@ -2057,3 +2057,6 @@#define PCI_DEVICE_ID_MICROGATE_USC 0x0010#define PCI_DEVICE_ID_MICROGATE_SCC 0x0020#define PCI_DEVICE_ID_MICROGATE_SCA 0x0030 + +#define PCI_VENDOR_ID_WORKBIT 0x1145 +#define PCI_DEVICE_ID_WORKBIT_CB 0xf021