[ink@jurassic.park.msu.ru: Re: [PATCH] Sort generic PCI fixups after specific ones]

From: Matthew Wilcox
Date: Wed Oct 06 2004 - 15:25:44 EST


Greg asked me to resend this patch and cc l-k, so here it is.

----- Forwarded message from Ivan Kokshaysky <ink@xxxxxxxxxxxxxxxxxxxx> -----

Date: Sat, 2 Oct 2004 01:48:17 +0400
From: Ivan Kokshaysky <ink@xxxxxxxxxxxxxxxxxxxx>
To: Matthew Wilcox <matthew@xxxxxx>
Cc: Greg KH <greg@xxxxxxxxx>, Linus Torvalds <torvalds@xxxxxxxx>,
Andrew Morton <akpm@xxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxx,
linux-pci@xxxxxxxxxxxxxxxxxxxxxxxx, parisc-linux@xxxxxxxxxxxxxxxx
Subject: Re: [PATCH] Sort generic PCI fixups after specific ones
User-Agent: Mutt/1.2.5i
In-Reply-To: <20040930174155.GT16153@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>; from matthew@xxxxxx on Thu, Sep 30, 2004 at 06:41:55PM +0100
X-Spam-Score: 0.0 (/)

On Thu, Sep 30, 2004 at 06:41:55PM +0100, Matthew Wilcox wrote:
> Allow prioritising PCI fixups. "How it works" is covered in the comment
> in pci.h. The patch to superio.c may well only apply with fuzz to the
> current Linux tree; I include it only to show an example.

No, you missed my point.
What we need is yet another PCI fixup *pass*, not prioritizing fixups
inside *one* pass - see appended patch (compile tested only).
Speaking of IDE: generally, you cannot switch PCI IDE controller
from "compatible" to native mode in the "fixup header" pass, as at this
point PCI BARs are *already* probed, and in compatible mode a lot of IDE
controllers have some bogus (often read-only) values in the BARs 0-3.
That's why quirk_ide_bases() fixup exists in the first place.
So that parisc superio fix works by mere luck - probably because IDE
BARs 0-3 are readable/writable even in legacy mode on this particular
controller.

Ivan.

--- 2.6/include/asm-generic/vmlinux.lds.h Mon Sep 27 14:38:14 2004
+++ linux/include/asm-generic/vmlinux.lds.h Fri Oct 1 15:24:30 2004
@@ -18,6 +18,9 @@
\
/* PCI quirks */ \
.pci_fixup : AT(ADDR(.pci_fixup) - LOAD_OFFSET) { \
+ VMLINUX_SYMBOL(__start_pci_fixups_early) = .; \
+ *(.pci_fixup_early) \
+ VMLINUX_SYMBOL(__end_pci_fixups_early) = .; \
VMLINUX_SYMBOL(__start_pci_fixups_header) = .; \
*(.pci_fixup_header) \
VMLINUX_SYMBOL(__end_pci_fixups_header) = .; \
--- 2.6/include/linux/pci.h Mon Sep 27 14:38:23 2004
+++ linux/include/linux/pci.h Fri Oct 1 16:16:51 2004
@@ -1006,20 +1006,29 @@ struct pci_fixup {
};

enum pci_fixup_pass {
- pci_fixup_header, /* Called immediately after reading configuration header */
+ pci_fixup_early, /* Called immediately after reading
+ device/vendor IDs and class code */
+ pci_fixup_header, /* Called after reading the entire
+ configuration header (including BARs) */
pci_fixup_final, /* Final phase of device fixups */
};

/* Anonymous variables would be nice... */
-#define DECLARE_PCI_FIXUP_HEADER(vendor, device, hook) \
- static struct pci_fixup __pci_fixup_##vendor##device##hook __attribute_used__ \
- __attribute__((__section__(".pci_fixup_header"))) = { \
- vendor, device, hook };
+#define DECLARE_PCI_FIXUP_SECTION(section, name, vendor, device, hook) \
+ static struct pci_fixup __pci_fixup_##name __attribute_used__ \
+ __attribute__((__section__( #section ))) = { vendor, device, hook };

-#define DECLARE_PCI_FIXUP_FINAL(vendor, device, hook) \
- static struct pci_fixup __pci_fixup_##vendor##device##hook __attribute_used__ \
- __attribute__((__section__(".pci_fixup_final"))) = { \
- vendor, device, hook };
+#define DECLARE_PCI_FIXUP_EARLY(vendor, device, hook) \
+ DECLARE_PCI_FIXUP_SECTION(.pci_fixup_early, \
+ vendor##device##hook, vendor, device, hook)
+
+#define DECLARE_PCI_FIXUP_HEADER(vendor, device, hook) \
+ DECLARE_PCI_FIXUP_SECTION(.pci_fixup_header, \
+ vendor##device##hook, vendor, device, hook)
+
+#define DECLARE_PCI_FIXUP_FINAL(vendor, device, hook) \
+ DECLARE_PCI_FIXUP_SECTION(.pci_fixup_final, \
+ vendor##device##hook, vendor, device, hook)

void pci_fixup_device(enum pci_fixup_pass pass, struct pci_dev *dev);

--- 2.6/drivers/pci/quirks.c Mon Sep 27 14:37:45 2004
+++ linux/drivers/pci/quirks.c Fri Oct 1 16:24:16 2004
@@ -999,6 +999,8 @@ static void pci_do_fixups(struct pci_dev
}
}

+extern struct pci_fixup __start_pci_fixups_early[];
+extern struct pci_fixup __end_pci_fixups_early[];
extern struct pci_fixup __start_pci_fixups_header[];
extern struct pci_fixup __end_pci_fixups_header[];
extern struct pci_fixup __start_pci_fixups_final[];
@@ -1009,6 +1011,11 @@ void pci_fixup_device(enum pci_fixup_pas
struct pci_fixup *start, *end;

switch(pass) {
+ case pci_fixup_early:
+ start = __start_pci_fixups_early;
+ end = __end_pci_fixups_early;
+ break;
+
case pci_fixup_header:
start = __start_pci_fixups_header;
end = __end_pci_fixups_header;
--- 2.6/drivers/pci/probe.c Mon Sep 27 14:37:45 2004
+++ linux/drivers/pci/probe.c Fri Oct 1 15:30:00 2004
@@ -475,6 +475,9 @@ static int pci_setup_device(struct pci_d
/* "Unknown power state" */
dev->current_state = 4;

+ /* Early fixups, before probing the BARs */
+ pci_fixup_device(pci_fixup_early, dev);
+
switch (dev->hdr_type) { /* header type */
case PCI_HEADER_TYPE_NORMAL: /* standard header */
if (class == PCI_CLASS_BRIDGE_PCI)
--- 2.6/drivers/parisc/superio.c Mon Sep 27 14:37:45 2004
+++ linux/drivers/parisc/superio.c Fri Oct 1 15:32:20 2004
@@ -484,7 +484,7 @@ void superio_fixup_pci(struct pci_dev *p
pci_read_config_byte(pdev, PCI_CLASS_PROG, &prog);
printk("PCI: Enabled native mode for NS87415 (pif=0x%x)\n", prog);
}
-DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NS, PCI_DEVICE_ID_NS_87415, superio_fixup_pci);
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_NS, PCI_DEVICE_ID_NS_87415, superio_fixup_pci);

/* Because of a defect in Super I/O, all reads of the PCI DMA status
* registers, IDE status register and the IDE select register need to be


----- End forwarded message -----

--
"Next the statesmen will invent cheap lies, putting the blame upon
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince
himself that the war is just, and will thank God for the better sleep
he enjoys after this process of grotesque self-deception." -- Mark Twain
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/