ultrastor.c is a bit-rot

Inspecting ultrastor.c it is clear to me that this was never used for a loooooooooong time. Not since a PC has more then 2^24 bit of memory. Let me explain below.

Now I'm not saying it should be fixed. I'm saying that it should be dumped in the account that it is not used by any one and that it does not work.

Why it never worked?
~~~~~~~~~~~~~~~~~~~~~

The driver's header says it supports 3 cards

* 14F - ISA first-party DMA HA with floppy support and WD1003 emulation.
* 24F - EISA Bus Master HA with floppy support and WD1003 emulation.
* 34F - VL-Bus Bus Master HA with floppy support (no WD1003 emulation).

But Kconfig only specifies ISA. I'm not sure what a VL-Bus is.

now the driver defines a static array of structures like this:

	struct {
	  ...
	
	  struct mscp mscp[ULTRASTOR_MAX_CMDS];
	} config = {0};

and allocates a struct mscp in .queuecommand like this:

	    my_mscp = &config.mscp[mscp_index];

it will go on preparing this my_mscp structure including stuffing some mapped pointers. Lets put that aside for now.
At the very end it will pass this my_mscp structure to the card's firmware like this:

	    /* Store pointer in OGM address bytes */
	outl(isa_virt_to_bus(my_mscp), config.ogm_address);

Now this is one hell of a smart ISA card. But putting this aside.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

Quoted 10 lines · Boaz Harrosh

Inspecting ultrastor.c it is clear to me that this was never used for a loooooooooong time. Not since a PC has more then 2^24 bit of memory. Let me explain below.

Now I'm not saying it should be fixed. I'm saying that it should be dumped in the account that it is not used by any one and that it does not work.

Why it never worked?
~~~~~~~~~~~~~~~~~~~~~

Quoted 7 lines

The driver's header says it supports 3 cards

* 14F - ISA first-party DMA HA with floppy support and WD1003 emulation.
* 24F - EISA Bus Master HA with floppy support and WD1003 emulation.
* 34F - VL-Bus Bus Master HA with floppy support (no WD1003 emulation).

But Kconfig only specifies ISA. I'm not sure what a VL-Bus is.

VL is vesa local ... it was an ISA like graphics bus that was fast and could reach > 16MB.

Quoted 16 lines

now the driver defines a static array of structures like this:

	struct {
	  ...
	
	  struct mscp mscp[ULTRASTOR_MAX_CMDS];
	} config = {0};

and allocates a struct mscp in .queuecommand like this:

	    my_mscp = &config.mscp[mscp_index];

it will go on preparing this my_mscp structure including stuffing some mapped pointers. Lets put that aside for now.
At the very end it will pass this my_mscp structure to the card's firmware like this:

Quoted 7 lines

	    /* Store pointer in OGM address bytes */
	outl(isa_virt_to_bus(my_mscp), config.ogm_address);

Now this is one hell of a smart ISA card. But putting this aside.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

It will definitely work for EISA and VL bus. I think if you analyse the placement of kernel data segments for compiled in drivers, it might also work for ISA too, since I think the pfn will be low enough. It should fail as a module not just because the area will be out of range for ISA, but also because the module data segment is in vmalloc space, so the virt_to_bus assumptions of contiguity could be violated.

James

Quoted 53 lines · James Bottomley

Quoted 10 lines · Boaz Harrosh

Inspecting ultrastor.c it is clear to me that this was never used for a loooooooooong time. Not since a PC has more then 2^24 bit of memory. Let me explain below.

Now I'm not saying it should be fixed. I'm saying that it should be dumped in the account that it is not used by any one and that it does not work.

Why it never worked?
~~~~~~~~~~~~~~~~~~~~~

Quoted 7 lines

The driver's header says it supports 3 cards

* 14F - ISA first-party DMA HA with floppy support and WD1003 emulation.
* 24F - EISA Bus Master HA with floppy support and WD1003 emulation.
* 34F - VL-Bus Bus Master HA with floppy support (no WD1003 emulation).

But Kconfig only specifies ISA. I'm not sure what a VL-Bus is.

VL is vesa local ... it was an ISA like graphics bus that was fast and could reach > 16MB.

Quoted 16 lines

now the driver defines a static array of structures like this:

	struct {
	  ...
	
	  struct mscp mscp[ULTRASTOR_MAX_CMDS];
	} config = {0};

and allocates a struct mscp in .queuecommand like this:

	    my_mscp = &config.mscp[mscp_index];

it will go on preparing this my_mscp structure including stuffing some mapped pointers. Lets put that aside for now.
At the very end it will pass this my_mscp structure to the card's firmware like this:

Quoted 7 lines

	    /* Store pointer in OGM address bytes */
	outl(isa_virt_to_bus(my_mscp), config.ogm_address);

Now this is one hell of a smart ISA card. But putting this aside.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

It will definitely work for EISA and VL bus. I think if you analyse the placement of kernel data segments for compiled in drivers, it might also work for ISA too, since I think the pfn will be low enough. It should fail as a module not just because the area will be out of range for ISA, but also because the module data segment is in vmalloc space, so the virt_to_bus assumptions of contiguity could be violated.

Quoted 2 lines

James

So what is the verdict? is it removed? marked broken for ISA?

can I safely say that unchecked_isa_dma can be removed?

Quoted 52 lines · Boaz Harrosh

Quoted 51 lines · James Bottomley

Quoted 10 lines · Boaz Harrosh

Inspecting ultrastor.c it is clear to me that this was never used for a loooooooooong time. Not since a PC has more then 2^24 bit of memory. Let me explain below.

Now I'm not saying it should be fixed. I'm saying that it should be dumped in the account that it is not used by any one and that it does not work.

Why it never worked?
~~~~~~~~~~~~~~~~~~~~~

Quoted 7 lines

The driver's header says it supports 3 cards

* 14F - ISA first-party DMA HA with floppy support and WD1003 emulation.
* 24F - EISA Bus Master HA with floppy support and WD1003 emulation.
* 34F - VL-Bus Bus Master HA with floppy support (no WD1003 emulation).

But Kconfig only specifies ISA. I'm not sure what a VL-Bus is.

VL is vesa local ... it was an ISA like graphics bus that was fast and could reach > 16MB.

Quoted 16 lines

now the driver defines a static array of structures like this:

	struct {
	  ...
	
	  struct mscp mscp[ULTRASTOR_MAX_CMDS];
	} config = {0};

and allocates a struct mscp in .queuecommand like this:

	    my_mscp = &config.mscp[mscp_index];

it will go on preparing this my_mscp structure including stuffing some mapped pointers. Lets put that aside for now.
At the very end it will pass this my_mscp structure to the card's firmware like this:

Quoted 7 lines

	    /* Store pointer in OGM address bytes */
	outl(isa_virt_to_bus(my_mscp), config.ogm_address);

Now this is one hell of a smart ISA card. But putting this aside.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

It will definitely work for EISA and VL bus. I think if you analyse the placement of kernel data segments for compiled in drivers, it might also work for ISA too, since I think the pfn will be low enough. It should fail as a module not just because the area will be out of range for ISA, but also because the module data segment is in vmalloc space, so the virt_to_bus assumptions of contiguity could be violated.

Quoted 3 lines

Quoted 2 lines

James

So what is the verdict? is it removed? marked broken for ISA?

It's probably obvious enough to apply the best straight line fix.

Quoted 1 line

can I safely say that unchecked_isa_dma can be removed?

No ... ISA definitely requires it.

James

In Hebrew we say: "You make me drink Kerosene".

An "obvious enough to apply the best straight line fix" submitted below:

I say dump it, it's unused.

Boaz
---
From: Boaz Harrosh <[email protected]>
Date: Mon, 17 Mar 2008 18:40:03 +0200
Subject: [PATCH] ultrastor: Fix for ISA DMA allocation

"obvious enough to apply the best straight line fix" submitted below.

Signed-off-by Boaz Harrosh <[email protected]>

CC Andi Kleen <[email protected]>

---
 drivers/scsi/ultrastor.c |   24 +++++++++++++++++++++---
 1 files changed, 21 insertions(+), 3 deletions(-)

drivers/scsi/ultrastor.c +3 -2

index f385dce..04441eb 100644
@@ -255,8 +255,9 @@ static struct ultrastor_config
   unsigned long mscp_free;
 #endif
   volatile unsigned char aborted[ULTRASTOR_MAX_CMDS];
-  struct mscp mscp[ULTRASTOR_MAX_CMDS];
-} config = {0};
+  struct mscp *mscp;
+  dma_addr_t dma;
+} config;
 
 /* Set this to 1 to reset the SCSI bus on error.  */
 static int ultrastor_bus_reset;

@@ -646,12 +647,29 @@ static int ultrastor_24f_detect(struct scsi_host_template * tpnt)
 
 static int ultrastor_detect(struct scsi_host_template * tpnt)

The driver's header says it supports 3 cards

* 14F - ISA first-party DMA HA with floppy support and WD1003 emulation.
* 24F - EISA Bus Master HA with floppy support and WD1003 emulation.
* 34F - VL-Bus Bus Master HA with floppy support (no WD1003 emulation).

But Kconfig only specifies ISA. I'm not sure what a VL-Bus is.

now the driver defines a static array of structures like this:

	struct {
	  ...
	
	  struct mscp mscp[ULTRASTOR_MAX_CMDS];
	} config = {0};

and allocates a struct mscp in .queuecommand like this:

	    my_mscp = &config.mscp[mscp_index];

it will go on preparing this my_mscp structure including stuffing some mapped pointers. Lets put that aside for now.
At the very end it will pass this my_mscp structure to the card's firmware like this:

	    /* Store pointer in OGM address bytes */
	outl(isa_virt_to_bus(my_mscp), config.ogm_address);

Now this is one hell of a smart ISA card. But putting this aside.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

(Also none of the emails in this file are valid)

VESA Local Bus. It was (in some sense) the predecessor of AGP. We treat it like ISA inside the kernel. On x86, EISA depends on ISA, so the dependency, while wrong, does not affect any x86 users who have an EISA card.

Quoted 5 lines · Matthew Wilcox

VESA Local Bus. It was (in some sense) the predecessor of AGP. We treat it like ISA inside the kernel. On x86, EISA depends on ISA, so the dependency, while wrong, does not affect any x86 users who have an EISA card.

Well, one can say VLB was to ISA (a souped-up ISA bus with some of the worst limitations removed) what AGP is to PCI...

-hpa

Quoted 6 lines · H. Peter Anvin

Quoted 5 lines · Matthew Wilcox

VESA Local Bus. It was (in some sense) the predecessor of AGP. We treat it like ISA inside the kernel. On x86, EISA depends on ISA, so the dependency, while wrong, does not affect any x86 users who have an EISA card.

Quoted 3 lines

Well, one can say VLB was to ISA (a souped-up ISA bus with some of the worst limitations removed) what AGP is to PCI...

-hpa

Hmm interesting, so someone took the VGA thing and made a storage device for it. Did that ever happen with AGP? Any AGP scsi cards. I guess I can Google for it.

Quoted 3 lines

Hmm interesting, so someone took the VGA thing and made a storage device for it. Did that ever happen with AGP? Any AGP scsi cards. I guess I can Google for it.

A lot less likely, IMO. Unlike AGP-PCI, VLB slots were compatible with ISA cards, and the electricals were less complex, so several motherboard manufacturers built boards with 3 or more VESA slots (I think I saw a board with 6 at one point.)

AGP electricals pretty much require that it be point to point.

... not that VLB didn't suck badly with several devices attached at the same time. I would be rather surprised if that board would work if fully populated with VLB cards, all in active use. IOW, a likely explanation is that it had a warning along the lines of "use of more than <number> of VLB cards at the same time may use instability" buried in documentation and proud "6 VLB slots!!!" touted by marketing...

I can't find VLB specs online, but IIRC 3 had been the limit and anything past that had been very much out of spec and likely to screw you.

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?

if the machine has more then 2^24 of memory. Then this will never work, right? or I'm missing it completely?