A moment ago I compiled 2.5.8 for a non-PCI 486.
Below a diff that makes it compile (and work),
modulo other patches that have been posted already.
diff -u --recursive --new-file ../linux-2.5.8/linux/drivers/ide/ide-proc.c ./linux/drivers/ide/ide-proc.c
--- ../linux-2.5.8/linux/drivers/ide/ide-proc.c Tue Apr 16 00:42:05 2002
+++ ./linux/drivers/ide/ide-proc.c Sun Apr 21 21:32:10 2002
@@ -422,7 +422,6 @@
static void create_proc_ide_drives(struct ata_channel *hwif)
{
int d;
- struct proc_dir_entry *ent;
struct proc_dir_entry *parent = hwif->proc;
char name[64];
diff -u --recursive --new-file ../linux-2.5.8/linux/drivers/ide/ide.c ./linux/drivers/ide/ide.c
--- ../linux-2.5.8/linux/drivers/ide/ide.c Tue Apr 16 00:42:05 2002
+++ ./linux/drivers/ide/ide.c Sun Apr 21 21:44:35 2002
@@ -2701,7 +2701,11 @@
void ide_teardown_commandlist(ide_drive_t *drive)
{
- struct pci_dev *pdev= drive->channel->pci_dev;
+#ifdef CONFIG_BLK_DEV_IDEPCI
+ struct pci_dev *pdev = drive->channel->pci_dev;
+#else
+ struct pci_dev *pdev = NULL;
+#endif
struct list_head *entry;
list_for_each(entry, &drive->free_req) {
@@ -2709,14 +2713,19 @@
list_del(&ar->ar_queue);
kfree(ar->ar_sg_table);
- pci_free_consistent(pdev, PRD_SEGMENTS * PRD_BYTES, ar->ar_dmatable_cpu, ar->ar_dmatable);
+ pci_free_consistent(pdev, PRD_SEGMENTS * PRD_BYTES,
+ ar->ar_dmatable_cpu, ar->ar_dmatable);
kfree(ar);
}
}
int ide_build_commandlist(ide_drive_t *drive)
{
- struct pci_dev *pdev= drive->channel->pci_dev;
+#ifdef CONFIG_BLK_DEV_IDEPCI
+ struct pci_dev *pdev = drive->channel->pci_dev;
+#else
+ struct pci_dev *pdev = NULL;
+#endif
struct ata_request *ar;
ide_tag_info_t *tcq;
int i, err;
diff -u --recursive --new-file ../linux-2.5.8/linux/include/asm-i386/ide.h ./linux/include/asm-i386/ide.h
--- ../linux-2.5.8/linux/include/asm-i386/ide.h Fri Mar 29 12:39:13 2002
+++ ./linux/include/asm-i386/ide.h Sun Apr 21 21:26:52 2002
@@ -75,6 +75,7 @@
static __inline__ void ide_init_default_hwifs(void)
{
#ifndef CONFIG_BLK_DEV_IDEPCI
+ extern int ide_register_hw(hw_regs_t *hw, struct ata_channel **hwifp);
hw_regs_t hw;
int index;
Of course this is not a suggestion to put in those #ifdef's,
this is just to point out that pci_dev is undefined when
CONFIG_BLK_DEV_IDEPCI is not set. You might consider putting
something like
#ifdef CONFIG_BLK_DEV_IDEPCI
#define PCI_DEV(drive) (drive->channel->pci_dev)
#else
#define PCI_DEV(drive) (NULL)
#endif
in some header file. Or, alternatively, just always define
the field pci_dev in this struct, wasting 4 bytes and avoiding
the #ifdef's altogether. In case you do that, soon the entire
CONFIG_BLK_DEV_IDEPCI becomes superfluous.
Andries
[email protected] wrote:
> A moment ago I compiled 2.5.8 for a non-PCI 486.
> Below a diff that makes it compile (and work),
> modulo other patches that have been posted already.
>
> diff -u --recursive --new-file ../linux-2.5.8/linux/drivers/ide/ide-proc.c ./linux/drivers/ide/ide-proc.c
> --- ../linux-2.5.8/linux/drivers/ide/ide-proc.c Tue Apr 16 00:42:05 2002
> +++ ./linux/drivers/ide/ide-proc.c Sun Apr 21 21:32:10 2002
> @@ -422,7 +422,6 @@
> static void create_proc_ide_drives(struct ata_channel *hwif)
> {
> int d;
> - struct proc_dir_entry *ent;
> struct proc_dir_entry *parent = hwif->proc;
> char name[64];
>
> diff -u --recursive --new-file ../linux-2.5.8/linux/drivers/ide/ide.c ./linux/drivers/ide/ide.c
> --- ../linux-2.5.8/linux/drivers/ide/ide.c Tue Apr 16 00:42:05 2002
> +++ ./linux/drivers/ide/ide.c Sun Apr 21 21:44:35 2002
> @@ -2701,7 +2701,11 @@
>
> void ide_teardown_commandlist(ide_drive_t *drive)
> {
> - struct pci_dev *pdev= drive->channel->pci_dev;
> +#ifdef CONFIG_BLK_DEV_IDEPCI
> + struct pci_dev *pdev = drive->channel->pci_dev;
> +#else
> + struct pci_dev *pdev = NULL;
> +#endif
> struct list_head *entry;
>
> list_for_each(entry, &drive->free_req) {
> @@ -2709,14 +2713,19 @@
>
> list_del(&ar->ar_queue);
> kfree(ar->ar_sg_table);
> - pci_free_consistent(pdev, PRD_SEGMENTS * PRD_BYTES, ar->ar_dmatable_cpu, ar->ar_dmatable);
> + pci_free_consistent(pdev, PRD_SEGMENTS * PRD_BYTES,
> + ar->ar_dmatable_cpu, ar->ar_dmatable);
> kfree(ar);
> }
> }
>
> int ide_build_commandlist(ide_drive_t *drive)
> {
> - struct pci_dev *pdev= drive->channel->pci_dev;
> +#ifdef CONFIG_BLK_DEV_IDEPCI
> + struct pci_dev *pdev = drive->channel->pci_dev;
> +#else
> + struct pci_dev *pdev = NULL;
> +#endif
> struct ata_request *ar;
> ide_tag_info_t *tcq;
> int i, err;
> diff -u --recursive --new-file ../linux-2.5.8/linux/include/asm-i386/ide.h ./linux/include/asm-i386/ide.h
> --- ../linux-2.5.8/linux/include/asm-i386/ide.h Fri Mar 29 12:39:13 2002
> +++ ./linux/include/asm-i386/ide.h Sun Apr 21 21:26:52 2002
> @@ -75,6 +75,7 @@
> static __inline__ void ide_init_default_hwifs(void)
> {
> #ifndef CONFIG_BLK_DEV_IDEPCI
> + extern int ide_register_hw(hw_regs_t *hw, struct ata_channel **hwifp);
> hw_regs_t hw;
> int index;
>
> Of course this is not a suggestion to put in those #ifdef's,
> this is just to point out that pci_dev is undefined when
> CONFIG_BLK_DEV_IDEPCI is not set. You might consider putting
> something like
>
> #ifdef CONFIG_BLK_DEV_IDEPCI
> #define PCI_DEV(drive) (drive->channel->pci_dev)
> #else
> #define PCI_DEV(drive) (NULL)
> #endif
Well this suggestion is indeed worth consideration.
However Linus is lagging right now a bit in pre patch
generatoin, becouse he apparently spends too much time
on meaningless discussions ;-).