Skip to contentSkip to conversation

Patch to make 2.4.13 compile on s390

PATCH Pete Zaitcev 2 participants 3 messages Oct 29, 2001

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Someone broke s390 and s390x, it did not even compile. Do we actually have a maintainer for that thing? Or do we blindly apply whatever IBM bestows?

MAINTAINERS lists some e-mails and lists for s390, but I did not receive any replies from those people. Must have been eaten by evil Lotus or something...

-- Pete

Makefile +0 -0
diff -ur -X dontdiff linux-2.4.13-0.2/Makefile linux-2.4.13-0.2-t1/Makefile
arch/s390/config.in +1 -0
diff -ur -X dontdiff linux-2.4.13-0.2/arch/s390/config.in linux-2.4.13-0.2-t1/arch/s390/config.in
@@ -10,6 +10,7 @@
 define_bool CONFIG_RWSEM_GENERIC_SPINLOCK y
 define_bool CONFIG_RWSEM_XCHGADD_ALGORITHM n
 define_bool CONFIG_GENERIC_BUST_SPINLOCK n
+define_bool CONFIG_GENERIC_ISA_DMA y
 
 mainmenu_name "Linux Kernel Configuration"
 define_bool CONFIG_ARCH_S390 y
arch/s390/mm/fault.c +10 -2
diff -ur -X dontdiff linux-2.4.13-0.2/arch/s390/mm/fault.c linux-2.4.13-0.2-t1/arch/s390/mm/fault.c
@@ -30,6 +30,7 @@
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
 #include <asm/hardirq.h>
+#include <asm/setup.h>
 
 #ifdef CONFIG_SYSCTL
 extern int sysctl_userprocess_debug;
@@ -38,7 +39,9 @@
 extern void die(const char *,struct pt_regs *,long);
 static void force_sigsegv(struct task_struct *tsk, int code, void *address);
 
+#if 0  /* #ifdef CONFIG_HWC_CONSOLE */	/* Ask IBM to fix XXX */
 extern spinlock_t timerlist_lock;
+#endif
 
 /*
  * Unlock any spinlocks which will prevent us from getting the
@@ -47,13 +50,18 @@
  */
 void bust_spinlocks(int yes)
 {
-	spin_lock_init(&timerlist_lock);
+#if 0  /* #ifdef CONFIG_HWC_CONSOLE */	/* Ask IBM to fix XXX */
+	spin_lock_init(&timerlist_lock);	/* Does not link... */
+#endif
 	if (yes) {
 		oops_in_progress = 1;
 	} else {
 		int loglevel_save = console_loglevel;
 		oops_in_progress = 0;
-		console_unblank();
+#ifdef CONFIG_HWC_CONSOLE
+		if (CONSOLE_IS_HWC)
+			hwc_console_unblank();
+#endif
 		/*
 		 * OK, the message is on the console.  Now we call printk()
 		 * without oops_in_progress set so that printk will give klogd
fs/partitions/ibm.c +2 -2
diff -ur -X dontdiff linux-2.4.13-0.2/fs/partitions/ibm.c linux-2.4.13-0.2-t1/fs/partitions/ibm.c
@@ -123,7 +123,7 @@
 					    GFP_KERNEL);
 	if ( geo == NULL )
 		return 0;
-	if (ioctl_by_bdev(bdev, HDIO_GETGEO, (unsigned long)geo);
+	if (ioctl_by_bdev(bdev, HDIO_GETGEO, (unsigned long)geo))
 		return 0;
 	blocksize = hardsect_size[MAJOR(dev)][MINOR(dev)];
 	if ( blocksize <= 0 ) {
@@ -131,7 +131,7 @@
 	}
 	blocksize >>= 9;
 	
-	data = read_dev_sector(bdev, inode->label_block*blocksize, &sect);
+	data = read_dev_sector(bdev, info->label_block*blocksize, &sect);
 	if (!data)
 		return 0;
 
#1 lore raw
Quoted 2 lines

Do we actually have a maintainer for that thing?
Or do we blindly apply whatever IBM bestows?

Sometimes generic fixes break stuff

arch/s390/config.in line 13

Quoted 11 lines

diff -ur -X dontdiff linux-2.4.13-0.2/Makefile linux-2.4.13-0.2-t1/Makefile

Makefile +0 -0

diff -ur -X dontdiff linux-2.4.13-0.2/arch/s390/config.in linux-2.4.13-0.2-t1/arch/s390/config.in

arch/s390/config.in +1 -0
@@ -10,6 +10,7 @@
 define_bool CONFIG_RWSEM_GENERIC_SPINLOCK y
 define_bool CONFIG_RWSEM_XCHGADD_ALGORITHM n
 define_bool CONFIG_GENERIC_BUST_SPINLOCK n
+define_bool CONFIG_GENERIC_ISA_DMA y

Please tell me where you find an S/390 mainframe with ISA bus. I think you may have fixed the effect not the cause.

#2 lore raw
In reply to Pete Zaitcev
Quoted 9 lines

Date: Mon, 29 Oct 2001 10:52:46 +0000 (GMT)
From: Alan Cox <[email protected]>

arch/s390/config.in line 13

Quoted 3 lines
arch/s390/config.in +1 -0
+define_bool CONFIG_GENERIC_ISA_DMA y

Please tell me where you find an S/390 mainframe with ISA bus. I think you may have fixed the effect not the cause.

This picks empty placeholders for request_dma, free_dma that kernel/dma.c provides. Before, dma.o always was compiled regardless of CONFIG_GENERIC_ISA_DMA.

-- Pete