* [patch 01/10] cio: change device sense procedure to work with pav aliases
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 02/10] magic sysrq: check for in_atomic before doing an console_unblank Martin Schwidefsky
` (8 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Peter Oberparleiter, Martin Schwidefsky
[-- Attachment #1: 001-cio-sense.diff --]
[-- Type: text/plain, Size: 3178 bytes --]
From: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Modify the sense id channel program to allow device sensing of pav
alias devices which belong to a base device with ungrouped paths.
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/device_id.c | 45 ++++++++-----------------------------------
1 file changed, 9 insertions(+), 36 deletions(-)
diff -urpN linux-2.6/drivers/s390/cio/device_id.c linux-2.6-patched/drivers/s390/cio/device_id.c
--- linux-2.6/drivers/s390/cio/device_id.c 2007-10-09 22:31:38.000000000 +0200
+++ linux-2.6-patched/drivers/s390/cio/device_id.c 2007-11-16 14:27:37.000000000 +0100
@@ -113,19 +113,10 @@ __ccw_device_sense_id_start(struct ccw_d
{
struct subchannel *sch;
struct ccw1 *ccw;
- int ret;
sch = to_subchannel(cdev->dev.parent);
/* Setup sense channel program. */
ccw = cdev->private->iccws;
- if (sch->schib.pmcw.pim != 0x80) {
- /* more than one path installed. */
- ccw->cmd_code = CCW_CMD_SUSPEND_RECONN;
- ccw->cda = 0;
- ccw->count = 0;
- ccw->flags = CCW_FLAG_SLI | CCW_FLAG_CC;
- ccw++;
- }
ccw->cmd_code = CCW_CMD_SENSE_ID;
ccw->cda = (__u32) __pa (&cdev->private->senseid);
ccw->count = sizeof (struct senseid);
@@ -133,25 +124,9 @@ __ccw_device_sense_id_start(struct ccw_d
/* Reset device status. */
memset(&cdev->private->irb, 0, sizeof(struct irb));
+ cdev->private->flags.intretry = 0;
- /* Try on every path. */
- ret = -ENODEV;
- while (cdev->private->imask != 0) {
- if ((sch->opm & cdev->private->imask) != 0 &&
- cdev->private->iretry > 0) {
- cdev->private->iretry--;
- /* Reset internal retry indication. */
- cdev->private->flags.intretry = 0;
- ret = cio_start (sch, cdev->private->iccws,
- cdev->private->imask);
- /* ret is 0, -EBUSY, -EACCES or -ENODEV */
- if (ret != -EACCES)
- return ret;
- }
- cdev->private->imask >>= 1;
- cdev->private->iretry = 5;
- }
- return ret;
+ return cio_start(sch, ccw, LPM_ANYPATH);
}
void
@@ -161,8 +136,7 @@ ccw_device_sense_id_start(struct ccw_dev
memset (&cdev->private->senseid, 0, sizeof (struct senseid));
cdev->private->senseid.cu_type = 0xFFFF;
- cdev->private->imask = 0x80;
- cdev->private->iretry = 5;
+ cdev->private->iretry = 3;
ret = __ccw_device_sense_id_start(cdev);
if (ret && ret != -EBUSY)
ccw_device_sense_id_done(cdev, ret);
@@ -278,14 +252,13 @@ ccw_device_sense_id_irq(struct ccw_devic
ccw_device_sense_id_done(cdev, ret);
break;
case -EACCES: /* channel is not operational. */
- sch->lpm &= ~cdev->private->imask;
- cdev->private->imask >>= 1;
- cdev->private->iretry = 5;
- /* fall through. */
case -EAGAIN: /* try again. */
- ret = __ccw_device_sense_id_start(cdev);
- if (ret == 0 || ret == -EBUSY)
- break;
+ cdev->private->iretry--;
+ if (cdev->private->iretry > 0) {
+ ret = __ccw_device_sense_id_start(cdev);
+ if (ret == 0 || ret == -EBUSY)
+ break;
+ }
/* fall through. */
default: /* Sense ID failed. Try asking VM. */
if (MACHINE_IS_VM) {
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 02/10] magic sysrq: check for in_atomic before doing an console_unblank
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
2007-11-16 14:24 ` [patch 01/10] cio: change device sense procedure to work with pav aliases Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 03/10] Fix irq tracing and lockdep_sys_exit calls Martin Schwidefsky
` (7 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Christian Borntraeger, Martin Schwidefsky
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: 002-console-unblank.diff --]
[-- Type: text/plain, Size: 2310 bytes --]
From: Christian Borntraeger <borntraeger@de.ibm.com>
When doing an magic sysrq reboot on s390 the following bug message
appears:
SysRq : Resetting
BUG: sleeping function called from invalid context at include/asm/semaphore.h:61
in_atomic():1, irqs_disabled():0
07000000004002a8 000000000fe6bc48 0000000000000002 0000000000000000
000000000fe6bce8 000000000fe6bc60 000000000fe6bc60 000000000012a79a
0000000000000000 07000000004002a8 0000000000000006 0000000000000000
0000000000000000 000000000fe6bc48 000000000000000d 000000000fe6bcb8
00000000004000c8 0000000000103234 000000000fe6bc48 000000000fe6bc90
Call Trace:
(¬<00000000001031b2>| show_trace+0x12e/0x148)
¬<000000000011ffca>| __might_sleep+0x10a/0x118
¬<0000000000129fba>| acquire_console_sem+0x92/0xf4
¬<000000000012a2ca>| console_unblank+0xc2/0xc8
¬<0000000000107bb4>| machine_restart+0x54/0x6c
¬<000000000028e806>| sysrq_handle_reboot+0x26/0x30
¬<000000000028e52a>| __handle_sysrq+0xa6/0x180
¬<0000000000140134>| run_workqueue+0xcc/0x18c
¬<000000000014029a>| worker_thread+0xa6/0x108
¬<00000000001458e4>| kthread+0x64/0x9c
¬<0000000000106f0e>| kernel_thread_starter+0x6/0xc
¬<0000000000106f08>| kernel_thread_starter+0x0/0xc
The only reason for doing a console_unblank on s390 is to flush the
log buffer. We have to check for in_atomic before doing a
console_unblank as the console is otherwise filled with an unrelated
bug message.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/setup.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff -urpN linux-2.6/arch/s390/kernel/setup.c linux-2.6-patched/arch/s390/kernel/setup.c
--- linux-2.6/arch/s390/kernel/setup.c 2007-10-09 22:31:38.000000000 +0200
+++ linux-2.6-patched/arch/s390/kernel/setup.c 2007-11-16 14:27:37.000000000 +0100
@@ -347,7 +347,7 @@ void (*_machine_power_off)(void) = do_ma
void machine_restart(char *command)
{
- if (!in_interrupt() || oops_in_progress)
+ if ((!in_interrupt() && !in_atomic()) || oops_in_progress)
/*
* Only unblank the console if we are called in enabled
* context or a bust_spinlocks cleared the way for us.
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 03/10] Fix irq tracing and lockdep_sys_exit calls.
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
2007-11-16 14:24 ` [patch 01/10] cio: change device sense procedure to work with pav aliases Martin Schwidefsky
2007-11-16 14:24 ` [patch 02/10] magic sysrq: check for in_atomic before doing an console_unblank Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 04/10] cmm: remove unused binary sysctls Martin Schwidefsky
` (6 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 003-lockdep.diff --]
[-- Type: text/plain, Size: 19554 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Current support for TRACE_IRQFLAGS and lockdep_sys_exit is broken.
IRQ flag tracing is broken for program checks. Even worse is that
the newly introduced calls to lockdep_sys_exit are in the critical
section code which is not supposed to call any C functions. In
addition the checks if locks are still held are also done when
returning to kernel code which is broken as well.
Fix all this by disabling interrupts and machine checks at the
exit paths and then do the appropriate checks and calls.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/entry.S | 109 +++++++++++++++++++++++++++++++--------------
arch/s390/kernel/entry64.S | 106 ++++++++++++++++++++++++++++++-------------
arch/s390/kernel/setup.c | 4 +
include/asm-s390/system.h | 5 ++
4 files changed, 158 insertions(+), 66 deletions(-)
diff -urpN linux-2.6/arch/s390/kernel/entry64.S linux-2.6-patched/arch/s390/kernel/entry64.S
--- linux-2.6/arch/s390/kernel/entry64.S 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/entry64.S 2007-11-16 14:27:38.000000000 +0100
@@ -67,12 +67,28 @@ _TIF_WORK_INT = (_TIF_SIGPENDING | _TIF_
brasl %r14,trace_hardirqs_off
.endm
- .macro LOCKDEP_SYS_EXIT
- brasl %r14,lockdep_sys_exit
+ .macro TRACE_IRQS_CHECK
+ tm SP_PSW(%r15),0x03 # irqs enabled?
+ jz 0f
+ brasl %r14,trace_hardirqs_on
+ j 1f
+0: brasl %r14,trace_hardirqs_off
+1:
.endm
#else
#define TRACE_IRQS_ON
#define TRACE_IRQS_OFF
+#define TRACE_IRQS_CHECK
+#endif
+
+#ifdef CONFIG_LOCKDEP
+ .macro LOCKDEP_SYS_EXIT
+ tm SP_PSW+1(%r15),0x01 # returning to user ?
+ jz 0f
+ brasl %r14,lockdep_sys_exit
+0:
+ .endm
+#else
#define LOCKDEP_SYS_EXIT
#endif
@@ -222,8 +238,6 @@ sysc_saveall:
llgh %r7,__LC_SVC_INT_CODE # get svc number from lowcore
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
sysc_vtime:
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- jz sysc_do_svc
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
sysc_stime:
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
@@ -257,19 +271,34 @@ sysc_noemu:
sysc_return:
tm SP_PSW+1(%r15),0x01 # returning to user ?
- jno sysc_leave
+ jno sysc_restore
tm __TI_flags+7(%r9),_TIF_WORK_SVC
jnz sysc_work # there is work to do (signals etc.)
+sysc_restore:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ larl %r1,sysc_restore_trace_psw
+ lpswe 0(%r1)
+sysc_restore_trace:
+ TRACE_IRQS_CHECK
LOCKDEP_SYS_EXIT
+#endif
sysc_leave:
RESTORE_ALL __LC_RETURN_PSW,1
+sysc_done:
+
+#ifdef CONFIG_TRACE_IRQFLAGS
+ .align 8
+ .globl sysc_restore_trace_psw
+sysc_restore_trace_psw:
+ .quad 0, sysc_restore_trace
+#endif
#
# recheck if there is more work to do
#
sysc_work_loop:
tm __TI_flags+7(%r9),_TIF_WORK_SVC
- jz sysc_leave # there is no work to do
+ jz sysc_restore # there is no work to do
#
# One of the work bits is on. Find out which one.
#
@@ -284,8 +313,8 @@ sysc_work:
jo sysc_restart
tm __TI_flags+7(%r9),_TIF_SINGLE_STEP
jo sysc_singlestep
- LOCKDEP_SYS_EXIT
- j sysc_leave
+ j sysc_restore
+sysc_work_done:
#
# _TIF_NEED_RESCHED is set, call schedule
@@ -445,6 +474,7 @@ pgm_check_handler:
pgm_no_vtime:
#endif
lg %r9,__LC_THREAD_INFO # load pointer to thread_info struct
+ TRACE_IRQS_OFF
lgf %r3,__LC_PGM_ILC # load program interruption code
lghi %r8,0x7f
ngr %r8,%r3
@@ -484,6 +514,7 @@ pgm_per_std:
pgm_no_vtime2:
#endif
lg %r9,__LC_THREAD_INFO # load pointer to thread_info struct
+ TRACE_IRQS_OFF
lg %r1,__TI_task(%r9)
tm SP_PSW+1(%r15),0x01 # kernel per event ?
jz kernel_per
@@ -504,12 +535,9 @@ pgm_svcper:
SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- jz pgm_no_vtime3
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
-pgm_no_vtime3:
#endif
llgh %r7,__LC_SVC_INT_CODE # get svc number from lowcore
lg %r9,__LC_THREAD_INFO # load pointer to thread_info struct
@@ -529,7 +557,7 @@ kernel_per:
lhi %r0,__LC_PGM_OLD_PSW
sth %r0,SP_TRAP(%r15) # set trap indication to pgm check
la %r2,SP_PTREGS(%r15) # address of register-save area
- larl %r14,sysc_leave # load adr. of system ret, no work
+ larl %r14,sysc_restore # load adr. of system ret, no work
jg do_single_step # branch to do_single_step
/*
@@ -554,26 +582,38 @@ io_no_vtime:
TRACE_IRQS_OFF
la %r2,SP_PTREGS(%r15) # address of register-save area
brasl %r14,do_IRQ # call standard irq handler
- TRACE_IRQS_ON
-
io_return:
tm SP_PSW+1(%r15),0x01 # returning to user ?
#ifdef CONFIG_PREEMPT
jno io_preempt # no -> check for preemptive scheduling
#else
- jno io_leave # no-> skip resched & signal
+ jno io_restore # no-> skip resched & signal
#endif
tm __TI_flags+7(%r9),_TIF_WORK_INT
jnz io_work # there is work to do (signals etc.)
+io_restore:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ larl %r1,io_restore_trace_psw
+ lpswe 0(%r1)
+io_restore_trace:
+ TRACE_IRQS_CHECK
LOCKDEP_SYS_EXIT
+#endif
io_leave:
RESTORE_ALL __LC_RETURN_PSW,0
io_done:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ .align 8
+ .globl io_restore_trace_psw
+io_restore_trace_psw:
+ .quad 0, io_restore_trace
+#endif
+
#ifdef CONFIG_PREEMPT
io_preempt:
icm %r0,15,__TI_precount(%r9)
- jnz io_leave
+ jnz io_restore
# switch to kernel stack
lg %r1,SP_R15(%r15)
aghi %r1,-SP_SIZE
@@ -582,12 +622,14 @@ io_preempt:
lgr %r15,%r1
io_resume_loop:
tm __TI_flags+7(%r9),_TIF_NEED_RESCHED
- jno io_leave
+ jno io_restore
larl %r1,.Lc_pactive
mvc __TI_precount(4,%r9),0(%r1)
+ TRACE_IRQS_ON
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
brasl %r14,schedule # call schedule
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
xc __TI_precount(4,%r9),__TI_precount(%r9)
j io_resume_loop
#endif
@@ -613,37 +655,39 @@ io_work_loop:
jo io_reschedule
tm __TI_flags+7(%r9),(_TIF_SIGPENDING | _TIF_RESTORE_SIGMASK)
jnz io_sigpending
- LOCKDEP_SYS_EXIT
- j io_leave
+ j io_restore
+io_work_done:
#
# _TIF_MCCK_PENDING is set, call handler
#
io_mcck_pending:
- TRACE_IRQS_OFF
brasl %r14,s390_handle_mcck # TIF bit will be cleared by handler
- TRACE_IRQS_ON
j io_work_loop
#
# _TIF_NEED_RESCHED is set, call schedule
#
io_reschedule:
+ TRACE_IRQS_ON
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
brasl %r14,schedule # call scheduler
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
tm __TI_flags+7(%r9),_TIF_WORK_INT
- jz io_leave # there is no work to do
+ jz io_restore # there is no work to do
j io_work_loop
#
# _TIF_SIGPENDING or _TIF_RESTORE_SIGMASK is set, call do_signal
#
io_sigpending:
+ TRACE_IRQS_ON
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
la %r2,SP_PTREGS(%r15) # load pt_regs
brasl %r14,do_signal # call do_signal
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
j io_work_loop
/*
@@ -669,7 +713,6 @@ ext_no_vtime:
la %r2,SP_PTREGS(%r15) # address of register-save area
llgh %r3,__LC_EXT_INT_CODE # get interruption code
brasl %r14,do_extint
- TRACE_IRQS_ON
j io_return
__critical_end:
@@ -824,15 +867,15 @@ cleanup_table_system_call:
cleanup_table_sysc_return:
.quad sysc_return, sysc_leave
cleanup_table_sysc_leave:
- .quad sysc_leave, sysc_work_loop
+ .quad sysc_leave, sysc_done
cleanup_table_sysc_work_loop:
- .quad sysc_work_loop, sysc_reschedule
+ .quad sysc_work_loop, sysc_work_done
cleanup_table_io_return:
.quad io_return, io_leave
cleanup_table_io_leave:
.quad io_leave, io_done
cleanup_table_io_work_loop:
- .quad io_work_loop, io_mcck_pending
+ .quad io_work_loop, io_work_done
cleanup_critical:
clc 8(8,%r12),BASED(cleanup_table_system_call)
@@ -901,8 +944,6 @@ cleanup_system_call:
cleanup_vtime:
clc __LC_RETURN_PSW+8(8),BASED(cleanup_system_call_insn+24)
jhe cleanup_stime
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- jz cleanup_novtime
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
cleanup_stime:
clc __LC_RETURN_PSW+8(8),BASED(cleanup_system_call_insn+32)
@@ -910,7 +951,6 @@ cleanup_stime:
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
cleanup_update:
mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
-cleanup_novtime:
#endif
mvc __LC_RETURN_PSW+8(8),BASED(cleanup_table_system_call+8)
la %r12,__LC_RETURN_PSW
@@ -949,10 +989,10 @@ cleanup_sysc_leave:
2: la %r12,__LC_RETURN_PSW
br %r14
cleanup_sysc_leave_insn:
+ .quad sysc_done - 4
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- .quad sysc_leave + 16
+ .quad sysc_done - 8
#endif
- .quad sysc_leave + 12
cleanup_io_return:
mvc __LC_RETURN_PSW(8),0(%r12)
@@ -979,10 +1019,10 @@ cleanup_io_leave:
2: la %r12,__LC_RETURN_PSW
br %r14
cleanup_io_leave_insn:
+ .quad io_done - 4
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- .quad io_leave + 20
+ .quad io_done - 8
#endif
- .quad io_leave + 16
/*
* Integer constants
diff -urpN linux-2.6/arch/s390/kernel/entry.S linux-2.6-patched/arch/s390/kernel/entry.S
--- linux-2.6/arch/s390/kernel/entry.S 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/entry.S 2007-11-16 14:27:38.000000000 +0100
@@ -69,13 +69,31 @@ STACK_SIZE = 1 << STACK_SHIFT
basr %r14,%r1
.endm
- .macro LOCKDEP_SYS_EXIT
- l %r1,BASED(.Llockdep_sys_exit)
+ .macro TRACE_IRQS_CHECK
+ tm SP_PSW(%r15),0x03 # irqs enabled?
+ jz 0f
+ l %r1,BASED(.Ltrace_irq_on)
+ basr %r14,%r1
+ j 1f
+0: l %r1,BASED(.Ltrace_irq_off)
basr %r14,%r1
+1:
.endm
#else
#define TRACE_IRQS_ON
#define TRACE_IRQS_OFF
+#define TRACE_IRQS_CHECK
+#endif
+
+#ifdef CONFIG_LOCKDEP
+ .macro LOCKDEP_SYS_EXIT
+ tm SP_PSW+1(%r15),0x01 # returning to user ?
+ jz 0f
+ l %r1,BASED(.Llockdep_sys_exit)
+ basr %r14,%r1
+0:
+ .endm
+#else
#define LOCKDEP_SYS_EXIT
#endif
@@ -234,8 +252,6 @@ sysc_saveall:
lh %r7,0x8a # get svc number from lowcore
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
sysc_vtime:
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- bz BASED(sysc_do_svc)
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
sysc_stime:
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
@@ -263,19 +279,34 @@ sysc_do_restart:
sysc_return:
tm SP_PSW+1(%r15),0x01 # returning to user ?
- bno BASED(sysc_leave)
+ bno BASED(sysc_restore)
tm __TI_flags+3(%r9),_TIF_WORK_SVC
bnz BASED(sysc_work) # there is work to do (signals etc.)
+sysc_restore:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ la %r1,BASED(sysc_restore_trace_psw)
+ lpsw 0(%r1)
+sysc_restore_trace:
+ TRACE_IRQS_CHECK
LOCKDEP_SYS_EXIT
+#endif
sysc_leave:
RESTORE_ALL __LC_RETURN_PSW,1
+sysc_done:
+
+#ifdef CONFIG_TRACE_IRQFLAGS
+ .align 8
+ .globl sysc_restore_trace_psw
+sysc_restore_trace_psw:
+ .long 0, sysc_restore_trace + 0x80000000
+#endif
#
# recheck if there is more work to do
#
sysc_work_loop:
tm __TI_flags+3(%r9),_TIF_WORK_SVC
- bz BASED(sysc_leave) # there is no work to do
+ bz BASED(sysc_restore) # there is no work to do
#
# One of the work bits is on. Find out which one.
#
@@ -290,8 +321,8 @@ sysc_work:
bo BASED(sysc_restart)
tm __TI_flags+3(%r9),_TIF_SINGLE_STEP
bo BASED(sysc_singlestep)
- LOCKDEP_SYS_EXIT
- b BASED(sysc_leave)
+ b BASED(sysc_restore)
+sysc_work_done:
#
# _TIF_NEED_RESCHED is set, call schedule
@@ -458,6 +489,7 @@ pgm_check_handler:
pgm_no_vtime:
#endif
l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
+ TRACE_IRQS_OFF
l %r3,__LC_PGM_ILC # load program interruption code
la %r8,0x7f
nr %r8,%r3
@@ -497,6 +529,7 @@ pgm_per_std:
pgm_no_vtime2:
#endif
l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
+ TRACE_IRQS_OFF
l %r1,__TI_task(%r9)
mvc __THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID
mvc __THREAD_per+__PER_address(4,%r1),__LC_PER_ADDRESS
@@ -517,15 +550,13 @@ pgm_svcper:
SAVE_ALL_SYNC __LC_SVC_OLD_PSW,__LC_SAVE_AREA
CREATE_STACK_FRAME __LC_SVC_OLD_PSW,__LC_SAVE_AREA
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- bz BASED(pgm_no_vtime3)
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
-pgm_no_vtime3:
#endif
lh %r7,0x8a # get svc number from lowcore
l %r9,__LC_THREAD_INFO # load pointer to thread_info struct
+ TRACE_IRQS_OFF
l %r1,__TI_task(%r9)
mvc __THREAD_per+__PER_atmid(2,%r1),__LC_PER_ATMID
mvc __THREAD_per+__PER_address(4,%r1),__LC_PER_ADDRESS
@@ -542,7 +573,7 @@ kernel_per:
mvi SP_TRAP+1(%r15),0x28 # set trap indication to pgm check
la %r2,SP_PTREGS(%r15) # address of register-save area
l %r1,BASED(.Lhandle_per) # load adr. of per handler
- la %r14,BASED(sysc_leave) # load adr. of system return
+ la %r14,BASED(sysc_restore)# load adr. of system return
br %r1 # branch to do_single_step
/*
@@ -569,26 +600,38 @@ io_no_vtime:
l %r1,BASED(.Ldo_IRQ) # load address of do_IRQ
la %r2,SP_PTREGS(%r15) # address of register-save area
basr %r14,%r1 # branch to standard irq handler
- TRACE_IRQS_ON
-
io_return:
tm SP_PSW+1(%r15),0x01 # returning to user ?
#ifdef CONFIG_PREEMPT
bno BASED(io_preempt) # no -> check for preemptive scheduling
#else
- bno BASED(io_leave) # no-> skip resched & signal
+ bno BASED(io_restore) # no-> skip resched & signal
#endif
tm __TI_flags+3(%r9),_TIF_WORK_INT
bnz BASED(io_work) # there is work to do (signals etc.)
+io_restore:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ la %r1,BASED(io_restore_trace_psw)
+ lpsw 0(%r1)
+io_restore_trace:
+ TRACE_IRQS_CHECK
LOCKDEP_SYS_EXIT
+#endif
io_leave:
RESTORE_ALL __LC_RETURN_PSW,0
io_done:
+#ifdef CONFIG_TRACE_IRQFLAGS
+ .align 8
+ .globl io_restore_trace_psw
+io_restore_trace_psw:
+ .long 0, io_restore_trace + 0x80000000
+#endif
+
#ifdef CONFIG_PREEMPT
io_preempt:
icm %r0,15,__TI_precount(%r9)
- bnz BASED(io_leave)
+ bnz BASED(io_restore)
l %r1,SP_R15(%r15)
s %r1,BASED(.Lc_spsize)
mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15)
@@ -596,12 +639,14 @@ io_preempt:
lr %r15,%r1
io_resume_loop:
tm __TI_flags+3(%r9),_TIF_NEED_RESCHED
- bno BASED(io_leave)
+ bno BASED(io_restore)
mvc __TI_precount(4,%r9),BASED(.Lc_pactive)
+ TRACE_IRQS_ON
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
l %r1,BASED(.Lschedule)
basr %r14,%r1 # call schedule
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
xc __TI_precount(4,%r9),__TI_precount(%r9)
b BASED(io_resume_loop)
#endif
@@ -627,40 +672,42 @@ io_work_loop:
bo BASED(io_reschedule)
tm __TI_flags+3(%r9),(_TIF_SIGPENDING | _TIF_RESTORE_SIGMASK)
bnz BASED(io_sigpending)
- LOCKDEP_SYS_EXIT
- b BASED(io_leave)
+ b BASED(io_restore)
+io_work_done:
#
# _TIF_MCCK_PENDING is set, call handler
#
io_mcck_pending:
- TRACE_IRQS_OFF
l %r1,BASED(.Ls390_handle_mcck)
basr %r14,%r1 # TIF bit will be cleared by handler
- TRACE_IRQS_ON
b BASED(io_work_loop)
#
# _TIF_NEED_RESCHED is set, call schedule
#
io_reschedule:
+ TRACE_IRQS_ON
l %r1,BASED(.Lschedule)
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
basr %r14,%r1 # call scheduler
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
tm __TI_flags+3(%r9),_TIF_WORK_INT
- bz BASED(io_leave) # there is no work to do
+ bz BASED(io_restore) # there is no work to do
b BASED(io_work_loop)
#
# _TIF_SIGPENDING or _TIF_RESTORE_SIGMASK is set, call do_signal
#
io_sigpending:
+ TRACE_IRQS_ON
stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
la %r2,SP_PTREGS(%r15) # load pt_regs
l %r1,BASED(.Ldo_signal)
basr %r14,%r1 # call do_signal
stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
+ TRACE_IRQS_OFF
b BASED(io_work_loop)
/*
@@ -688,7 +735,6 @@ ext_no_vtime:
lh %r3,__LC_EXT_INT_CODE # get interruption code
l %r1,BASED(.Ldo_extint)
basr %r14,%r1
- TRACE_IRQS_ON
b BASED(io_return)
__critical_end:
@@ -853,15 +899,15 @@ cleanup_table_system_call:
cleanup_table_sysc_return:
.long sysc_return + 0x80000000, sysc_leave + 0x80000000
cleanup_table_sysc_leave:
- .long sysc_leave + 0x80000000, sysc_work_loop + 0x80000000
+ .long sysc_leave + 0x80000000, sysc_done + 0x80000000
cleanup_table_sysc_work_loop:
- .long sysc_work_loop + 0x80000000, sysc_reschedule + 0x80000000
+ .long sysc_work_loop + 0x80000000, sysc_work_done + 0x80000000
cleanup_table_io_return:
.long io_return + 0x80000000, io_leave + 0x80000000
cleanup_table_io_leave:
.long io_leave + 0x80000000, io_done + 0x80000000
cleanup_table_io_work_loop:
- .long io_work_loop + 0x80000000, io_mcck_pending + 0x80000000
+ .long io_work_loop + 0x80000000, io_work_done + 0x80000000
cleanup_critical:
clc 4(4,%r12),BASED(cleanup_table_system_call)
@@ -930,8 +976,6 @@ cleanup_system_call:
cleanup_vtime:
clc __LC_RETURN_PSW+4(4),BASED(cleanup_system_call_insn+12)
bhe BASED(cleanup_stime)
- tm SP_PSW+1(%r15),0x01 # interrupting from user ?
- bz BASED(cleanup_novtime)
UPDATE_VTIME __LC_EXIT_TIMER,__LC_SYNC_ENTER_TIMER,__LC_USER_TIMER
cleanup_stime:
clc __LC_RETURN_PSW+4(4),BASED(cleanup_system_call_insn+16)
@@ -939,7 +983,6 @@ cleanup_stime:
UPDATE_VTIME __LC_LAST_UPDATE_TIMER,__LC_EXIT_TIMER,__LC_SYSTEM_TIMER
cleanup_update:
mvc __LC_LAST_UPDATE_TIMER(8),__LC_SYNC_ENTER_TIMER
-cleanup_novtime:
#endif
mvc __LC_RETURN_PSW+4(4),BASED(cleanup_table_system_call+4)
la %r12,__LC_RETURN_PSW
@@ -978,10 +1021,10 @@ cleanup_sysc_leave:
2: la %r12,__LC_RETURN_PSW
br %r14
cleanup_sysc_leave_insn:
+ .long sysc_done - 4 + 0x80000000
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- .long sysc_leave + 14 + 0x80000000
+ .long sysc_done - 8 + 0x80000000
#endif
- .long sysc_leave + 10 + 0x80000000
cleanup_io_return:
mvc __LC_RETURN_PSW(4),0(%r12)
@@ -1008,10 +1051,10 @@ cleanup_io_leave:
2: la %r12,__LC_RETURN_PSW
br %r14
cleanup_io_leave_insn:
+ .long io_done - 4 + 0x80000000
#ifdef CONFIG_VIRT_CPU_ACCOUNTING
- .long io_leave + 18 + 0x80000000
+ .long io_done - 8 + 0x80000000
#endif
- .long io_leave + 14 + 0x80000000
/*
* Integer constants
diff -urpN linux-2.6/arch/s390/kernel/setup.c linux-2.6-patched/arch/s390/kernel/setup.c
--- linux-2.6/arch/s390/kernel/setup.c 2007-11-16 14:27:38.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/setup.c 2007-11-16 14:27:38.000000000 +0100
@@ -492,6 +492,10 @@ static void setup_addressing_mode(void)
printk("S390 address spaces switched, ");
set_amode_and_uaccess(PSW_ASC_PRIMARY, PSW32_ASC_PRIMARY);
}
+#ifdef CONFIG_TRACE_IRQFLAGS
+ sysc_restore_trace_psw.mask = psw_kernel_bits & ~PSW_MASK_MCHECK;
+ io_restore_trace_psw.mask = psw_kernel_bits & ~PSW_MASK_MCHECK;
+#endif
}
static void __init
diff -urpN linux-2.6/include/asm-s390/system.h linux-2.6-patched/include/asm-s390/system.h
--- linux-2.6/include/asm-s390/system.h 2007-11-16 14:27:34.000000000 +0100
+++ linux-2.6-patched/include/asm-s390/system.h 2007-11-16 14:27:38.000000000 +0100
@@ -388,6 +388,11 @@ extern void (*_machine_power_off)(void);
#define arch_align_stack(x) (x)
+#ifdef CONFIG_TRACE_IRQFLAGS
+extern psw_t sysc_restore_trace_psw;
+extern psw_t io_restore_trace_psw;
+#endif
+
#endif /* __KERNEL__ */
#endif
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 04/10] cmm: remove unused binary sysctls.
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (2 preceding siblings ...)
2007-11-16 14:24 ` [patch 03/10] Fix irq tracing and lockdep_sys_exit calls Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 05/10] appldata: " Martin Schwidefsky
` (5 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390
Cc: Christian Borntraeger, Heiko Carstens, Eric W. Biederman,
Andrew Morton, Martin Schwidefsky
[-- Attachment #1: 004-cmm.diff --]
[-- Type: text/plain, Size: 2436 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Remove binary sysctls that never worked due to missing strategy functions.
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/mm/cmm.c | 3 ---
include/linux/sysctl.h | 5 -----
kernel/sysctl_check.c | 3 ---
3 files changed, 11 deletions(-)
diff -urpN linux-2.6/arch/s390/mm/cmm.c linux-2.6-patched/arch/s390/mm/cmm.c
--- linux-2.6/arch/s390/mm/cmm.c 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/mm/cmm.c 2007-11-16 14:27:41.000000000 +0100
@@ -341,19 +341,16 @@ cmm_timeout_handler(ctl_table *ctl, int
static struct ctl_table cmm_table[] = {
{
- .ctl_name = VM_CMM_PAGES,
.procname = "cmm_pages",
.mode = 0644,
.proc_handler = &cmm_pages_handler,
},
{
- .ctl_name = VM_CMM_TIMED_PAGES,
.procname = "cmm_timed_pages",
.mode = 0644,
.proc_handler = &cmm_pages_handler,
},
{
- .ctl_name = VM_CMM_TIMEOUT,
.procname = "cmm_timeout",
.mode = 0644,
.proc_handler = &cmm_timeout_handler,
diff -urpN linux-2.6/include/linux/sysctl.h linux-2.6-patched/include/linux/sysctl.h
--- linux-2.6/include/linux/sysctl.h 2007-11-16 14:27:34.000000000 +0100
+++ linux-2.6-patched/include/linux/sysctl.h 2007-11-16 14:27:41.000000000 +0100
@@ -207,11 +207,6 @@ enum
VM_PANIC_ON_OOM=33, /* panic at out-of-memory */
VM_VDSO_ENABLED=34, /* map VDSO into new processes? */
VM_MIN_SLAB=35, /* Percent pages ignored by zone reclaim */
-
- /* s390 vm cmm sysctls */
- VM_CMM_PAGES=1111,
- VM_CMM_TIMED_PAGES=1112,
- VM_CMM_TIMEOUT=1113,
};
diff -urpN linux-2.6/kernel/sysctl_check.c linux-2.6-patched/kernel/sysctl_check.c
--- linux-2.6/kernel/sysctl_check.c 2007-11-16 14:27:35.000000000 +0100
+++ linux-2.6-patched/kernel/sysctl_check.c 2007-11-16 14:27:41.000000000 +0100
@@ -140,9 +140,6 @@ static struct trans_ctl_table trans_vm_t
{ VM_PANIC_ON_OOM, "panic_on_oom" },
{ VM_VDSO_ENABLED, "vdso_enabled" },
{ VM_MIN_SLAB, "min_slab_ratio" },
- { VM_CMM_PAGES, "cmm_pages" },
- { VM_CMM_TIMED_PAGES, "cmm_timed_pages" },
- { VM_CMM_TIMEOUT, "cmm_timeout" },
{}
};
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 05/10] appldata: remove unused binary sysctls.
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (3 preceding siblings ...)
2007-11-16 14:24 ` [patch 04/10] cmm: remove unused binary sysctls Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 06/10] Fix kernel preemption Martin Schwidefsky
` (4 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390
Cc: Eric W. Biederman, Christian Borntraeger, Gerald Schaefer,
Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 005-appldata.diff --]
[-- Type: text/plain, Size: 8946 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Remove binary sysctls that never worked due to missing strategy functions.
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Gerald Schaefer <geraldsc@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/appldata/appldata.h | 1
arch/s390/appldata/appldata_base.c | 74 ++++++----------------------------
arch/s390/appldata/appldata_mem.c | 1
arch/s390/appldata/appldata_net_sum.c | 1
arch/s390/appldata/appldata_os.c | 1
include/linux/sysctl.h | 1
kernel/sysctl_check.c | 11 -----
7 files changed, 15 insertions(+), 75 deletions(-)
diff -urpN linux-2.6/arch/s390/appldata/appldata_base.c linux-2.6-patched/arch/s390/appldata/appldata_base.c
--- linux-2.6/arch/s390/appldata/appldata_base.c 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/appldata/appldata_base.c 2007-11-16 14:27:42.000000000 +0100
@@ -53,29 +53,26 @@ static int appldata_interval_handler(ctl
static struct ctl_table_header *appldata_sysctl_header;
static struct ctl_table appldata_table[] = {
{
- .ctl_name = CTL_APPLDATA_TIMER,
.procname = "timer",
.mode = S_IRUGO | S_IWUSR,
.proc_handler = &appldata_timer_handler,
},
{
- .ctl_name = CTL_APPLDATA_INTERVAL,
.procname = "interval",
.mode = S_IRUGO | S_IWUSR,
.proc_handler = &appldata_interval_handler,
},
- { .ctl_name = 0 }
+ { },
};
static struct ctl_table appldata_dir_table[] = {
{
- .ctl_name = CTL_APPLDATA,
.procname = appldata_proc_name,
.maxlen = 0,
.mode = S_IRUGO | S_IXUGO,
.child = appldata_table,
},
- { .ctl_name = 0 }
+ { },
};
/*
@@ -441,75 +438,38 @@ out:
*/
int appldata_register_ops(struct appldata_ops *ops)
{
- struct list_head *lh;
- struct appldata_ops *tmp_ops;
- int i;
-
- i = 0;
+ if ((ops->size > APPLDATA_MAX_REC_SIZE) || (ops->size < 0))
+ return -EINVAL;
- if ((ops->size > APPLDATA_MAX_REC_SIZE) ||
- (ops->size < 0)){
- P_ERROR("Invalid size of %s record = %i, maximum = %i!\n",
- ops->name, ops->size, APPLDATA_MAX_REC_SIZE);
- return -ENOMEM;
- }
- if ((ops->ctl_nr == CTL_APPLDATA) ||
- (ops->ctl_nr == CTL_APPLDATA_TIMER) ||
- (ops->ctl_nr == CTL_APPLDATA_INTERVAL)) {
- P_ERROR("ctl_nr %i already in use!\n", ops->ctl_nr);
- return -EBUSY;
- }
- ops->ctl_table = kzalloc(4*sizeof(struct ctl_table), GFP_KERNEL);
- if (ops->ctl_table == NULL) {
- P_ERROR("Not enough memory for %s ctl_table!\n", ops->name);
+ ops->ctl_table = kzalloc(4 * sizeof(struct ctl_table), GFP_KERNEL);
+ if (!ops->ctl_table)
return -ENOMEM;
- }
spin_lock(&appldata_ops_lock);
- list_for_each(lh, &appldata_ops_list) {
- tmp_ops = list_entry(lh, struct appldata_ops, list);
- P_DEBUG("register_ops loop: %i) name = %s, ctl = %i\n",
- ++i, tmp_ops->name, tmp_ops->ctl_nr);
- P_DEBUG("Comparing %s (ctl %i) with %s (ctl %i)\n",
- tmp_ops->name, tmp_ops->ctl_nr, ops->name,
- ops->ctl_nr);
- if (strncmp(tmp_ops->name, ops->name,
- APPLDATA_PROC_NAME_LENGTH) == 0) {
- P_ERROR("Name \"%s\" already registered!\n", ops->name);
- kfree(ops->ctl_table);
- spin_unlock(&appldata_ops_lock);
- return -EBUSY;
- }
- if (tmp_ops->ctl_nr == ops->ctl_nr) {
- P_ERROR("ctl_nr %i already registered!\n", ops->ctl_nr);
- kfree(ops->ctl_table);
- spin_unlock(&appldata_ops_lock);
- return -EBUSY;
- }
- }
list_add(&ops->list, &appldata_ops_list);
spin_unlock(&appldata_ops_lock);
- ops->ctl_table[0].ctl_name = CTL_APPLDATA;
ops->ctl_table[0].procname = appldata_proc_name;
ops->ctl_table[0].maxlen = 0;
ops->ctl_table[0].mode = S_IRUGO | S_IXUGO;
ops->ctl_table[0].child = &ops->ctl_table[2];
- ops->ctl_table[1].ctl_name = 0;
-
- ops->ctl_table[2].ctl_name = ops->ctl_nr;
ops->ctl_table[2].procname = ops->name;
ops->ctl_table[2].mode = S_IRUGO | S_IWUSR;
ops->ctl_table[2].proc_handler = appldata_generic_handler;
ops->ctl_table[2].data = ops;
- ops->ctl_table[3].ctl_name = 0;
-
ops->sysctl_header = register_sysctl_table(ops->ctl_table);
-
+ if (!ops->sysctl_header)
+ goto out;
P_INFO("%s-ops registered!\n", ops->name);
return 0;
+out:
+ spin_lock(&appldata_ops_lock);
+ list_del(&ops->list);
+ spin_unlock(&appldata_ops_lock);
+ kfree(ops->ctl_table);
+ return -ENOMEM;
}
/*
@@ -519,15 +479,11 @@ int appldata_register_ops(struct appldat
*/
void appldata_unregister_ops(struct appldata_ops *ops)
{
- void *table;
spin_lock(&appldata_ops_lock);
list_del(&ops->list);
- /* at that point any incoming access will fail */
- table = ops->ctl_table;
- ops->ctl_table = NULL;
spin_unlock(&appldata_ops_lock);
unregister_sysctl_table(ops->sysctl_header);
- kfree(table);
+ kfree(ops->ctl_table);
P_INFO("%s-ops unregistered!\n", ops->name);
}
/********************** module-ops management <END> **************************/
diff -urpN linux-2.6/arch/s390/appldata/appldata.h linux-2.6-patched/arch/s390/appldata/appldata.h
--- linux-2.6/arch/s390/appldata/appldata.h 2007-10-09 22:31:38.000000000 +0200
+++ linux-2.6-patched/arch/s390/appldata/appldata.h 2007-11-16 14:27:42.000000000 +0100
@@ -45,7 +45,6 @@ struct appldata_ops {
int active; /* monitoring status */
/* fill in from here */
- unsigned int ctl_nr; /* sysctl ID */
char name[APPLDATA_PROC_NAME_LENGTH]; /* name of /proc fs node */
unsigned char record_nr; /* Record Nr. for Product ID */
void (*callback)(void *data); /* callback function */
diff -urpN linux-2.6/arch/s390/appldata/appldata_mem.c linux-2.6-patched/arch/s390/appldata/appldata_mem.c
--- linux-2.6/arch/s390/appldata/appldata_mem.c 2007-10-09 22:31:38.000000000 +0200
+++ linux-2.6-patched/arch/s390/appldata/appldata_mem.c 2007-11-16 14:27:42.000000000 +0100
@@ -147,7 +147,6 @@ static void appldata_get_mem_data(void *
static struct appldata_ops ops = {
- .ctl_nr = CTL_APPLDATA_MEM,
.name = "mem",
.record_nr = APPLDATA_RECORD_MEM_ID,
.size = sizeof(struct appldata_mem_data),
diff -urpN linux-2.6/arch/s390/appldata/appldata_net_sum.c linux-2.6-patched/arch/s390/appldata/appldata_net_sum.c
--- linux-2.6/arch/s390/appldata/appldata_net_sum.c 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/appldata/appldata_net_sum.c 2007-11-16 14:27:42.000000000 +0100
@@ -142,7 +142,6 @@ static void appldata_get_net_sum_data(vo
static struct appldata_ops ops = {
- .ctl_nr = CTL_APPLDATA_NET_SUM,
.name = "net_sum",
.record_nr = APPLDATA_RECORD_NET_SUM_ID,
.size = sizeof(struct appldata_net_sum_data),
diff -urpN linux-2.6/arch/s390/appldata/appldata_os.c linux-2.6-patched/arch/s390/appldata/appldata_os.c
--- linux-2.6/arch/s390/appldata/appldata_os.c 2007-10-09 22:31:38.000000000 +0200
+++ linux-2.6-patched/arch/s390/appldata/appldata_os.c 2007-11-16 14:27:42.000000000 +0100
@@ -82,7 +82,6 @@ struct appldata_os_data {
static struct appldata_os_data *appldata_os_data;
static struct appldata_ops ops = {
- .ctl_nr = CTL_APPLDATA_OS,
.name = "os",
.record_nr = APPLDATA_RECORD_OS_ID,
.owner = THIS_MODULE,
diff -urpN linux-2.6/include/linux/sysctl.h linux-2.6-patched/include/linux/sysctl.h
--- linux-2.6/include/linux/sysctl.h 2007-11-16 14:27:42.000000000 +0100
+++ linux-2.6-patched/include/linux/sysctl.h 2007-11-16 14:27:42.000000000 +0100
@@ -70,7 +70,6 @@ enum
CTL_ABI=9, /* Binary emulation */
CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
CTL_ARLAN=254, /* arlan wireless driver */
- CTL_APPLDATA=2120, /* s390 appldata */
CTL_S390DBF=5677, /* s390 debug */
CTL_SUNRPC=7249, /* sunrpc debug */
CTL_PM=9899, /* frv power management */
diff -urpN linux-2.6/kernel/sysctl_check.c linux-2.6-patched/kernel/sysctl_check.c
--- linux-2.6/kernel/sysctl_check.c 2007-11-16 14:27:42.000000000 +0100
+++ linux-2.6-patched/kernel/sysctl_check.c 2007-11-16 14:27:42.000000000 +0100
@@ -1216,16 +1216,6 @@ static struct trans_ctl_table trans_arla
{}
};
-static struct trans_ctl_table trans_appldata_table[] = {
- { CTL_APPLDATA_TIMER, "timer" },
- { CTL_APPLDATA_INTERVAL, "interval" },
- { CTL_APPLDATA_OS, "os" },
- { CTL_APPLDATA_NET_SUM, "net_sum" },
- { CTL_APPLDATA_MEM, "mem" },
- {}
-
-};
-
static struct trans_ctl_table trans_s390dbf_table[] = {
{ 5678 /* CTL_S390DBF_STOPPABLE */, "debug_stoppable" },
{ 5679 /* CTL_S390DBF_ACTIVE */, "debug_active" },
@@ -1270,7 +1260,6 @@ static struct trans_ctl_table trans_root
{ CTL_ABI, "abi" },
/* CTL_CPU not used */
{ CTL_ARLAN, "arlan", trans_arlan_table },
- { CTL_APPLDATA, "appldata", trans_appldata_table },
{ CTL_S390DBF, "s390dbf", trans_s390dbf_table },
{ CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
{ CTL_PM, "pm", trans_pm_table },
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 06/10] Fix kernel preemption.
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (4 preceding siblings ...)
2007-11-16 14:24 ` [patch 05/10] appldata: " Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 07/10] Optimize storage key handling for anonymous pages Martin Schwidefsky
` (3 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 006-preemption.diff --]
[-- Type: text/plain, Size: 2901 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
When returning from IRQ handling and TIF_NEED_RESCHED is set we must
call preempt_schedule_irq() instead of schedule().
Otherwise the BKL might be unlocked in schedule() and therfore
everything that relies on the BKL is broken.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/entry.S | 15 +++++----------
arch/s390/kernel/entry64.S | 12 ++----------
2 files changed, 7 insertions(+), 20 deletions(-)
diff -urpN linux-2.6/arch/s390/kernel/entry64.S linux-2.6-patched/arch/s390/kernel/entry64.S
--- linux-2.6/arch/s390/kernel/entry64.S 2007-11-16 14:27:41.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/entry64.S 2007-11-16 14:27:42.000000000 +0100
@@ -623,15 +623,8 @@ io_preempt:
io_resume_loop:
tm __TI_flags+7(%r9),_TIF_NEED_RESCHED
jno io_restore
- larl %r1,.Lc_pactive
- mvc __TI_precount(4,%r9),0(%r1)
- TRACE_IRQS_ON
- stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
- brasl %r14,schedule # call schedule
- stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
- TRACE_IRQS_OFF
- xc __TI_precount(4,%r9),__TI_precount(%r9)
- j io_resume_loop
+ larl %r14,io_resume_loop
+ jg preempt_schedule_irq
#endif
#
@@ -1029,7 +1022,6 @@ cleanup_io_leave_insn:
*/
.align 4
.Lconst:
-.Lc_pactive: .long PREEMPT_ACTIVE
.Lnr_syscalls: .long NR_syscalls
.L0x0130: .short 0x130
.L0x0140: .short 0x140
diff -urpN linux-2.6/arch/s390/kernel/entry.S linux-2.6-patched/arch/s390/kernel/entry.S
--- linux-2.6/arch/s390/kernel/entry.S 2007-11-16 14:27:41.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/entry.S 2007-11-16 14:27:42.000000000 +0100
@@ -640,15 +640,9 @@ io_preempt:
io_resume_loop:
tm __TI_flags+3(%r9),_TIF_NEED_RESCHED
bno BASED(io_restore)
- mvc __TI_precount(4,%r9),BASED(.Lc_pactive)
- TRACE_IRQS_ON
- stosm __SF_EMPTY(%r15),0x03 # reenable interrupts
- l %r1,BASED(.Lschedule)
- basr %r14,%r1 # call schedule
- stnsm __SF_EMPTY(%r15),0xfc # disable I/O and ext. interrupts
- TRACE_IRQS_OFF
- xc __TI_precount(4,%r9),__TI_precount(%r9)
- b BASED(io_resume_loop)
+ l %r1,BASED(.Lpreempt_schedule_irq)
+ la %r14,BASED(io_resume_loop)
+ br %r1 # call schedule
#endif
#
@@ -1062,7 +1056,6 @@ cleanup_io_leave_insn:
.align 4
.Lc_spsize: .long SP_SIZE
.Lc_overhead: .long STACK_FRAME_OVERHEAD
-.Lc_pactive: .long PREEMPT_ACTIVE
.Lnr_syscalls: .long NR_syscalls
.L0x018: .short 0x018
.L0x020: .short 0x020
@@ -1086,6 +1079,8 @@ cleanup_io_leave_insn:
.Lexecve_tail: .long execve_tail
.Ljump_table: .long pgm_check_table
.Lschedule: .long schedule
+.Lpreempt_schedule_irq:
+ .long preempt_schedule_irq
.Ltrace: .long syscall_trace
.Lschedtail: .long schedule_tail
.Lsysc_table: .long sys_call_table
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 07/10] Optimize storage key handling for anonymous pages
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (5 preceding siblings ...)
2007-11-16 14:24 ` [patch 06/10] Fix kernel preemption Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 08/10] Dont overwrite lowcores on smp_send_stop() Martin Schwidefsky
` (2 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Christian Borntraeger, Martin Schwidefsky
[-- Attachment #1: 007-dirty-bit.diff --]
[-- Type: text/plain, Size: 1601 bytes --]
From: Christian Borntraeger <borntraeger@de.ibm.com>
page_mkclean used to call page_clear_dirty for every given page. This
is different to all other architectures, where the dirty bit in the
PTEs is only resetted, if page_mapping() returns a non-NULL pointer.
We can move the page_test_dirty/page_clear_dirty sequence into the
2nd if to avoid unnecessary iske/sske sequences, which are expensive.
This change also helps kvm for s390 as the host must transfer the
dirty bit into the guest status bits. By moving the page_clear_dirty
operation into the 2nd if, the vm will only call page_clear_dirty
for pages where it walks the mapping anyway. There it calls
ptep_clear_flush for writable ptes, so we can transfer the dirty bit
to the guest.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
mm/rmap.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff -urpN linux-2.6/mm/rmap.c linux-2.6-patched/mm/rmap.c
--- linux-2.6/mm/rmap.c 2007-11-16 14:27:35.000000000 +0100
+++ linux-2.6-patched/mm/rmap.c 2007-11-16 14:27:43.000000000 +0100
@@ -471,11 +471,12 @@ int page_mkclean(struct page *page)
if (page_mapped(page)) {
struct address_space *mapping = page_mapping(page);
- if (mapping)
+ if (mapping) {
ret = page_mkclean_file(mapping, page);
- if (page_test_dirty(page)) {
- page_clear_dirty(page);
- ret = 1;
+ if (page_test_dirty(page)) {
+ page_clear_dirty(page);
+ ret = 1;
+ }
}
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 08/10] Dont overwrite lowcores on smp_send_stop().
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (6 preceding siblings ...)
2007-11-16 14:24 ` [patch 07/10] Optimize storage key handling for anonymous pages Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 09/10] Explicitly code allocpercpu calls in iucv Martin Schwidefsky
2007-11-16 14:24 ` [patch 10/10] Fix memory detection Martin Schwidefsky
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 008-lowcores.diff --]
[-- Type: text/plain, Size: 2706 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Don't perform a sigp store-status-at-address on smp_send_stop().
It will overwrite the lowcores of other cpus and destroys valueable
debug informations.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/smp.c | 56 ++++++-------------------------------------------
1 file changed, 7 insertions(+), 49 deletions(-)
diff -urpN linux-2.6/arch/s390/kernel/smp.c linux-2.6-patched/arch/s390/kernel/smp.c
--- linux-2.6/arch/s390/kernel/smp.c 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/smp.c 2007-11-16 14:27:44.000000000 +0100
@@ -193,72 +193,30 @@ int smp_call_function_single(int cpu, vo
}
EXPORT_SYMBOL(smp_call_function_single);
-static void do_send_stop(void)
+void smp_send_stop(void)
{
int cpu, rc;
- /* stop all processors */
- for_each_online_cpu(cpu) {
- if (cpu == smp_processor_id())
- continue;
- do {
- rc = signal_processor(cpu, sigp_stop);
- } while (rc == sigp_busy);
- }
-}
+ /* Disable all interrupts/machine checks */
+ __load_psw_mask(psw_kernel_bits & ~PSW_MASK_MCHECK);
-static void do_store_status(void)
-{
- int cpu, rc;
+ /* write magic number to zero page (absolute 0) */
+ lowcore_ptr[smp_processor_id()]->panic_magic = __PANIC_MAGIC;
- /* store status of all processors in their lowcores (real 0) */
+ /* stop all processors */
for_each_online_cpu(cpu) {
if (cpu == smp_processor_id())
continue;
do {
- rc = signal_processor_p(
- (__u32)(unsigned long) lowcore_ptr[cpu], cpu,
- sigp_store_status_at_address);
+ rc = signal_processor(cpu, sigp_stop);
} while (rc == sigp_busy);
- }
-}
-static void do_wait_for_stop(void)
-{
- int cpu;
-
- /* Wait for all other cpus to enter stopped state */
- for_each_online_cpu(cpu) {
- if (cpu == smp_processor_id())
- continue;
while (!smp_cpu_not_running(cpu))
cpu_relax();
}
}
/*
- * this function sends a 'stop' sigp to all other CPUs in the system.
- * it goes straight through.
- */
-void smp_send_stop(void)
-{
- /* Disable all interrupts/machine checks */
- __load_psw_mask(psw_kernel_bits & ~PSW_MASK_MCHECK);
-
- /* write magic number to zero page (absolute 0) */
- lowcore_ptr[smp_processor_id()]->panic_magic = __PANIC_MAGIC;
-
- /* stop other processors. */
- do_send_stop();
-
- /* wait until other processors are stopped */
- do_wait_for_stop();
-
- /* store status of other processors. */
- do_store_status();
-}
-
-/*
* Reboot, halt and power_off routines for SMP.
*/
void machine_restart_smp(char *__unused)
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 09/10] Explicitly code allocpercpu calls in iucv
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (7 preceding siblings ...)
2007-11-16 14:24 ` [patch 08/10] Dont overwrite lowcores on smp_send_stop() Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
2007-11-16 14:24 ` [patch 10/10] Fix memory detection Martin Schwidefsky
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Christoph Lameter, Martin Schwidefsky
[-- Attachment #1: 009-iucv.diff --]
[-- Type: text/plain, Size: 10184 bytes --]
From: Christoph Lameter <clameter@sgi.com>
The iucv is the only user of the various functions that are used to bring
parts of cpus up and down. Its the only allocpercpu user that will do
I/O on per cpu objects (which is difficult to do with virtually mapped memory).
And its the only use of allocpercpu where a GFP_DMA allocation is done.
Remove the allocpercpu calls from iucv and code the allocation and freeing
manually. After this patch it is possible to remove a large part of
the allocpercpu API.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
net/iucv/iucv.c | 107 +++++++++++++++++++++++++++++++-------------------------
1 file changed, 61 insertions(+), 46 deletions(-)
diff -urpN linux-2.6/net/iucv/iucv.c linux-2.6-patched/net/iucv/iucv.c
--- linux-2.6/net/iucv/iucv.c 2007-11-16 14:27:35.000000000 +0100
+++ linux-2.6-patched/net/iucv/iucv.c 2007-11-16 14:27:44.000000000 +0100
@@ -97,7 +97,7 @@ struct iucv_irq_list {
struct iucv_irq_data data;
};
-static struct iucv_irq_data *iucv_irq_data;
+static struct iucv_irq_data *iucv_irq_data[NR_CPUS];
static cpumask_t iucv_buffer_cpumask = CPU_MASK_NONE;
static cpumask_t iucv_irq_cpumask = CPU_MASK_NONE;
@@ -277,7 +277,7 @@ union iucv_param {
/*
* Anchor for per-cpu IUCV command parameter block.
*/
-static union iucv_param *iucv_param;
+static union iucv_param *iucv_param[NR_CPUS];
/**
* iucv_call_b2f0
@@ -356,7 +356,7 @@ static void iucv_allow_cpu(void *data)
* 0x10 - Flag to allow priority message completion interrupts
* 0x08 - Flag to allow IUCV control interrupts
*/
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[cpu];
memset(parm, 0, sizeof(union iucv_param));
parm->set_mask.ipmask = 0xf8;
iucv_call_b2f0(IUCV_SETMASK, parm);
@@ -377,7 +377,7 @@ static void iucv_block_cpu(void *data)
union iucv_param *parm;
/* Disable all iucv interrupts. */
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[cpu];
memset(parm, 0, sizeof(union iucv_param));
iucv_call_b2f0(IUCV_SETMASK, parm);
@@ -401,9 +401,9 @@ static void iucv_declare_cpu(void *data)
return;
/* Declare interrupt buffer. */
- parm = percpu_ptr(iucv_param, cpu);
+ parm = iucv_param[cpu];
memset(parm, 0, sizeof(union iucv_param));
- parm->db.ipbfadr1 = virt_to_phys(percpu_ptr(iucv_irq_data, cpu));
+ parm->db.ipbfadr1 = virt_to_phys(iucv_irq_data[cpu]);
rc = iucv_call_b2f0(IUCV_DECLARE_BUFFER, parm);
if (rc) {
char *err = "Unknown";
@@ -458,7 +458,7 @@ static void iucv_retrieve_cpu(void *data
iucv_block_cpu(NULL);
/* Retrieve interrupt buffer. */
- parm = percpu_ptr(iucv_param, cpu);
+ parm = iucv_param[cpu];
iucv_call_b2f0(IUCV_RETRIEVE_BUFFER, parm);
/* Clear indication that an iucv buffer exists for this cpu. */
@@ -558,22 +558,23 @@ static int __cpuinit iucv_cpu_notify(str
switch (action) {
case CPU_UP_PREPARE:
case CPU_UP_PREPARE_FROZEN:
- if (!percpu_populate(iucv_irq_data,
- sizeof(struct iucv_irq_data),
- GFP_KERNEL|GFP_DMA, cpu))
+ iucv_irq_data[cpu] = kmalloc_node(sizeof(struct iucv_irq_data),
+ GFP_KERNEL|GFP_DMA, cpu_to_node(cpu));
+ if (!iucv_irq_data[cpu])
return NOTIFY_BAD;
- if (!percpu_populate(iucv_param, sizeof(union iucv_param),
- GFP_KERNEL|GFP_DMA, cpu)) {
- percpu_depopulate(iucv_irq_data, cpu);
+ iucv_param[cpu] = kmalloc_node(sizeof(union iucv_param),
+ GFP_KERNEL|GFP_DMA, cpu_to_node(cpu));
+ if (!iucv_param[cpu])
return NOTIFY_BAD;
- }
break;
case CPU_UP_CANCELED:
case CPU_UP_CANCELED_FROZEN:
case CPU_DEAD:
case CPU_DEAD_FROZEN:
- percpu_depopulate(iucv_param, cpu);
- percpu_depopulate(iucv_irq_data, cpu);
+ kfree(iucv_param[cpu]);
+ iucv_param[cpu] = NULL;
+ kfree(iucv_irq_data[cpu]);
+ iucv_irq_data[cpu] = NULL;
break;
case CPU_ONLINE:
case CPU_ONLINE_FROZEN:
@@ -612,7 +613,7 @@ static int iucv_sever_pathid(u16 pathid,
{
union iucv_param *parm;
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (userdata)
memcpy(parm->ctrl.ipuser, userdata, sizeof(parm->ctrl.ipuser));
@@ -755,7 +756,7 @@ int iucv_path_accept(struct iucv_path *p
local_bh_disable();
/* Prepare parameter block. */
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
parm->ctrl.ippathid = path->pathid;
parm->ctrl.ipmsglim = path->msglim;
@@ -799,7 +800,7 @@ int iucv_path_connect(struct iucv_path *
BUG_ON(in_atomic());
spin_lock_bh(&iucv_table_lock);
iucv_cleanup_queue();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
parm->ctrl.ipmsglim = path->msglim;
parm->ctrl.ipflags1 = path->flags;
@@ -854,7 +855,7 @@ int iucv_path_quiesce(struct iucv_path *
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (userdata)
memcpy(parm->ctrl.ipuser, userdata, sizeof(parm->ctrl.ipuser));
@@ -881,7 +882,7 @@ int iucv_path_resume(struct iucv_path *p
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (userdata)
memcpy(parm->ctrl.ipuser, userdata, sizeof(parm->ctrl.ipuser));
@@ -936,7 +937,7 @@ int iucv_message_purge(struct iucv_path
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
parm->purge.ippathid = path->pathid;
parm->purge.ipmsgid = msg->id;
@@ -1003,7 +1004,7 @@ int iucv_message_receive(struct iucv_pat
}
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
parm->db.ipbfadr1 = (u32)(addr_t) buffer;
parm->db.ipbfln1f = (u32) size;
@@ -1040,7 +1041,7 @@ int iucv_message_reject(struct iucv_path
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
parm->db.ippathid = path->pathid;
parm->db.ipmsgid = msg->id;
@@ -1074,7 +1075,7 @@ int iucv_message_reply(struct iucv_path
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (flags & IUCV_IPRMDATA) {
parm->dpl.ippathid = path->pathid;
@@ -1118,7 +1119,7 @@ int iucv_message_send(struct iucv_path *
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (flags & IUCV_IPRMDATA) {
/* Message of 8 bytes can be placed into the parameter list. */
@@ -1172,7 +1173,7 @@ int iucv_message_send2way(struct iucv_pa
int rc;
local_bh_disable();
- parm = percpu_ptr(iucv_param, smp_processor_id());
+ parm = iucv_param[smp_processor_id()];
memset(parm, 0, sizeof(union iucv_param));
if (flags & IUCV_IPRMDATA) {
parm->dpl.ippathid = path->pathid;
@@ -1559,7 +1560,7 @@ static void iucv_external_interrupt(u16
struct iucv_irq_data *p;
struct iucv_irq_list *work;
- p = percpu_ptr(iucv_irq_data, smp_processor_id());
+ p = iucv_irq_data[smp_processor_id()];
if (p->ippathid >= iucv_max_pathid) {
printk(KERN_WARNING "iucv_do_int: Got interrupt with "
"pathid %d > max_connections (%ld)\n",
@@ -1598,6 +1599,7 @@ static void iucv_external_interrupt(u16
static int __init iucv_init(void)
{
int rc;
+ int cpu;
if (!MACHINE_IS_VM) {
rc = -EPROTONOSUPPORT;
@@ -1617,19 +1619,23 @@ static int __init iucv_init(void)
rc = PTR_ERR(iucv_root);
goto out_bus;
}
- /* Note: GFP_DMA used to get memory below 2G */
- iucv_irq_data = percpu_alloc(sizeof(struct iucv_irq_data),
- GFP_KERNEL|GFP_DMA);
- if (!iucv_irq_data) {
- rc = -ENOMEM;
- goto out_root;
- }
- /* Allocate parameter blocks. */
- iucv_param = percpu_alloc(sizeof(union iucv_param),
- GFP_KERNEL|GFP_DMA);
- if (!iucv_param) {
- rc = -ENOMEM;
- goto out_extint;
+
+ for_each_online_cpu(cpu) {
+ /* Note: GFP_DMA used to get memory below 2G */
+ iucv_irq_data[cpu] = kmalloc_node(sizeof(struct iucv_irq_data),
+ GFP_KERNEL|GFP_DMA, cpu_to_node(cpu));
+ if (!iucv_irq_data[cpu]) {
+ rc = -ENOMEM;
+ goto out_free;
+ }
+
+ /* Allocate parameter blocks. */
+ iucv_param[cpu] = kmalloc_node(sizeof(union iucv_param),
+ GFP_KERNEL|GFP_DMA, cpu_to_node(cpu));
+ if (!iucv_param[cpu]) {
+ rc = -ENOMEM;
+ goto out_free;
+ }
}
register_hotcpu_notifier(&iucv_cpu_notifier);
ASCEBC(iucv_error_no_listener, 16);
@@ -1638,9 +1644,13 @@ static int __init iucv_init(void)
iucv_available = 1;
return 0;
-out_extint:
- percpu_free(iucv_irq_data);
-out_root:
+out_free:
+ for_each_possible_cpu(cpu) {
+ kfree(iucv_param[cpu]);
+ iucv_param[cpu] = NULL;
+ kfree(iucv_irq_data[cpu]);
+ iucv_irq_data[cpu] = NULL;
+ }
s390_root_dev_unregister(iucv_root);
out_bus:
bus_unregister(&iucv_bus);
@@ -1658,6 +1668,7 @@ out:
static void __exit iucv_exit(void)
{
struct iucv_irq_list *p, *n;
+ int cpu;
spin_lock_irq(&iucv_queue_lock);
list_for_each_entry_safe(p, n, &iucv_task_queue, list)
@@ -1666,8 +1677,12 @@ static void __exit iucv_exit(void)
kfree(p);
spin_unlock_irq(&iucv_queue_lock);
unregister_hotcpu_notifier(&iucv_cpu_notifier);
- percpu_free(iucv_param);
- percpu_free(iucv_irq_data);
+ for_each_possible_cpu(cpu) {
+ kfree(iucv_param[cpu]);
+ iucv_param[cpu] = NULL;
+ kfree(iucv_irq_data[cpu]);
+ iucv_irq_data[cpu] = NULL;
+ }
s390_root_dev_unregister(iucv_root);
bus_unregister(&iucv_bus);
unregister_external_interrupt(0x4000, iucv_external_interrupt);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread* [patch 10/10] Fix memory detection.
2007-11-16 14:24 [patch 00/10] 2.6.24-rc s390 patch queue Martin Schwidefsky
` (8 preceding siblings ...)
2007-11-16 14:24 ` [patch 09/10] Explicitly code allocpercpu calls in iucv Martin Schwidefsky
@ 2007-11-16 14:24 ` Martin Schwidefsky
9 siblings, 0 replies; 11+ messages in thread
From: Martin Schwidefsky @ 2007-11-16 14:24 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 010-memsize.diff --]
[-- Type: text/plain, Size: 1105 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Before we're getting short on memory detection fixes here is the next
one: if neither sclp nor diag260 report the storage size the detection
loop will return immediately without detecting anything. Fix this by
breaking the detection loop only if the memory end is known.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/early.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff -urpN linux-2.6/arch/s390/kernel/early.c linux-2.6-patched/arch/s390/kernel/early.c
--- linux-2.6/arch/s390/kernel/early.c 2007-11-16 14:27:30.000000000 +0100
+++ linux-2.6-patched/arch/s390/kernel/early.c 2007-11-16 14:27:45.000000000 +0100
@@ -200,7 +200,7 @@ static noinline __init void find_memory_
cc = __tprot(addr);
while (cc == old_cc) {
addr += CHUNK_INCR;
- if (addr >= memsize)
+ if (memsize && addr >= memsize)
break;
#ifndef CONFIG_64BIT
if (addr == ADDR2G)
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 11+ messages in thread