Select this option, if you want to share the text segment of the
Linux kernel between different VM guests. This reduces memory
usage with lots of guests but greatly increases kernel size.
+ Also if a kernel was IPL'ed from a shared segment the kexec system
+ call will not work.
You should only select this option if you know what you are
doing and want to exploit this feature.
.long .Lduct # cr2: dispatchable unit control table
.long 0 # cr3: instruction authorization
.long 0 # cr4: instruction authorization
- .long 0xffffffff # cr5: primary-aste origin
+ .long .Lduct # cr5: primary-aste origin
.long 0 # cr6: I/O interrupts
.long 0 # cr7: secondary space segment table
.long 0 # cr8: access registers translation
.long 0 # cr13: home space segment table
.long 0xc0000000 # cr14: machine check handling off
.long 0 # cr15: linkage stack operations
-.Lduct: .long 0,0,0,0,0,0,0,0
- .long 0,0,0,0,0,0,0,0
.Lpcfpu:.long 0x00080000,0x80000000 + .Lchkfpu
.Lpccsp:.long 0x00080000,0x80000000 + .Lchkcsp
.Lpcmvpg:.long 0x00080000,0x80000000 + .Lchkmvpg
.Linittu: .long init_thread_union
.Lstartup_init:
.long startup_init
+ .align 64
+.Lduct: .long 0,0,0,0,.Lduald,0,0,0
+ .long 0,0,0,0,0,0,0,0
+ .align 128
+.Lduald:.rept 8
+ .long 0x80000000,0,0,0 # invalid access-list entries
+ .endr
.org 0x12000
.globl _ehead
.quad .Lduct # cr2: dispatchable unit control table
.quad 0 # cr3: instruction authorization
.quad 0 # cr4: instruction authorization
- .quad 0xffffffffffffffff # cr5: primary-aste origin
+ .quad .Lduct # cr5: primary-aste origin
.quad 0 # cr6: I/O interrupts
.quad 0 # cr7: secondary space segment table
.quad 0 # cr8: access registers translation
.quad 0 # cr13: home space segment table
.quad 0xc0000000 # cr14: machine check handling off
.quad 0 # cr15: linkage stack operations
-.Lduct: .long 0,0,0,0,0,0,0,0
- .long 0,0,0,0,0,0,0,0
.Lpcmsk:.quad 0x0000000180000000
.L4malign:.quad 0xffffffffffc00000
.Lscan2g:.quad 0x80000000 + 0x20000 - 8 # 2GB + 128K - 8
.Lnop: .long 0x07000700
.Lparmaddr:
.quad PARMAREA
+ .align 64
+.Lduct: .long 0,0,0,0,.Lduald,0,0,0
+ .long 0,0,0,0,0,0,0,0
+ .align 128
+.Lduald:.rept 8
+ .long 0x80000000,0,0,0 # invalid access-list entries
+ .endr
.org 0x12000
.globl _ehead
reset->fn();
}
-extern __u32 dump_prefix_page;
+u32 dump_prefix_page;
void s390_reset_system(void)
{
lc->panic_stack = S390_lowcore.panic_stack;
/* Save prefix page address for dump case */
- dump_prefix_page = (unsigned long) lc;
+ dump_prefix_page = (u32)(unsigned long) lc;
/* Disable prefixing */
set_prefix(0);
}
p = get_kprobe(addr);
- if (!p) {
- if (*addr != BREAKPOINT_INSTRUCTION) {
- /*
- * The breakpoint instruction was removed right
- * after we hit it. Another cpu has removed
- * either a probepoint or a debugger breakpoint
- * at this address. In either case, no further
- * handling of this interrupt is appropriate.
- *
- */
- ret = 1;
- }
- /* Not one of ours: let kernel handle it */
+ if (!p)
+ /*
+ * No kprobe at this address. The fault has not been
+ * caused by a kprobe breakpoint. The race of breakpoint
+ * vs. kprobe remove does not exist because on s390 we
+ * use stop_machine_run to arm/disarm the breakpoints.
+ */
goto no_kprobe;
- }
kcb->kprobe_status = KPROBE_HIT_ACTIVE;
set_current_kprobe(p, regs, kcb);
#include <asm/system.h>
#include <asm/smp.h>
#include <asm/reset.h>
+#include <asm/ipl.h>
typedef void (*relocate_kernel_t)(kimage_entry_t *, unsigned long);
{
void *reboot_code_buffer;
+ /* Can't replace kernel image since it is read-only. */
+ if (ipl_flags & IPL_NSS_VALID)
+ return -ENOSYS;
+
/* We don't support anything but the default image type for now. */
if (image->type != KEXEC_TYPE_DEFAULT)
return -EINVAL;
#include <asm/lowcore.h>
+#
+# do_reipl_asm
+# Parameter: r2 = schid of reipl device
+#
.globl do_reipl_asm
do_reipl_asm: basr %r13,0
.Lpg0: lpsw .Lnewpsw-.Lpg0(%r13)
stm %r0,%r15,__LC_GPREGS_SAVE_AREA
stctl %c0,%c15,__LC_CREGS_SAVE_AREA
stam %a0,%a15,__LC_AREGS_SAVE_AREA
- mvc __LC_PREFIX_SAVE_AREA(4),dump_prefix_page-.Lpg0(%r13)
+ l %r10,.Ldump_pfx-.Lpg0(%r13)
+ mvc __LC_PREFIX_SAVE_AREA(4),0(%r10)
stckc .Lclkcmp-.Lpg0(%r13)
mvc __LC_CLOCK_COMP_SAVE_AREA(8),.Lclkcmp-.Lpg0(%r13)
stpt __LC_CPU_TIMER_SAVE_AREA
st %r13, __LC_PSW_SAVE_AREA+4
-
lctl %c6,%c6,.Lall-.Lpg0(%r13)
lr %r1,%r2
mvc __LC_PGM_NEW_PSW(8),.Lpcnew-.Lpg0(%r13)
.align 8
.Lclkcmp: .quad 0x0000000000000000
.Lall: .long 0xff000000
+.Ldump_pfx: .long dump_prefix_page
.align 8
.Lnewpsw: .long 0x00080000,0x80000000+.Lpg1
.Lpcnew: .long 0x00080000,0x80000000+.Lecs
.long 0x00000000,0x00000000
.long 0x00000000,0x00000000
.long 0x00000000,0x00000000
- .globl dump_prefix_page
-dump_prefix_page:
- .long 0x00000000
-
*/
#include <asm/lowcore.h>
+
+#
+# do_reipl_asm
+# Parameter: r2 = schid of reipl device
+#
+
.globl do_reipl_asm
do_reipl_asm: basr %r13,0
.Lpg0: lpswe .Lnewpsw-.Lpg0(%r13)
stg %r0,__LC_GPREGS_SAVE_AREA-0x1000+8(%r1)
stctg %c0,%c15,__LC_CREGS_SAVE_AREA-0x1000(%r1)
stam %a0,%a15,__LC_AREGS_SAVE_AREA-0x1000(%r1)
- mvc __LC_PREFIX_SAVE_AREA-0x1000(4,%r1),dump_prefix_page-.Lpg0(%r13)
+ lg %r10,.Ldump_pfx-.Lpg0(%r13)
+ mvc __LC_PREFIX_SAVE_AREA-0x1000(4,%r1),0(%r10)
stfpc __LC_FP_CREG_SAVE_AREA-0x1000(%r1)
stckc .Lclkcmp-.Lpg0(%r13)
mvc __LC_CLOCK_COMP_SAVE_AREA-0x1000(8,%r1),.Lclkcmp-.Lpg0(%r13)
.align 8
.Lclkcmp: .quad 0x0000000000000000
.Lall: .quad 0x00000000ff000000
+.Ldump_pfx: .quad dump_prefix_page
.Lregsave: .quad 0x0000000000000000
.align 16
/*
.long 0x00000000,0x00000000
.long 0x00000000,0x00000000
.long 0x00000000,0x00000000
- .globl dump_prefix_page
-dump_prefix_page:
- .long 0x00000000
int cpu, local = 0;
/*
- * Can deadlock when interrupts are disabled or if in wrong context,
- * caller must disable preemption
+ * Can deadlock when interrupts are disabled or if in wrong context.
*/
- WARN_ON(irqs_disabled() || in_irq() || preemptible());
+ WARN_ON(irqs_disabled() || in_irq());
/*
* Check for local function call. We have to have the same call order
* Run a function on all other CPUs.
*
* You must not call this function with disabled interrupts or from a
- * hardware interrupt handler. Must be called with preemption disabled.
- * You may call it from a bottom half.
+ * hardware interrupt handler. You may call it from a bottom half.
*/
int smp_call_function(void (*func) (void *info), void *info, int nonatomic,
int wait)
{
cpumask_t map;
+ preempt_disable();
map = cpu_online_map;
cpu_clear(smp_processor_id(), map);
__smp_call_function_map(func, info, nonatomic, wait, map);
+ preempt_enable();
return 0;
}
EXPORT_SYMBOL(smp_call_function);
* Run a function on one processor.
*
* You must not call this function with disabled interrupts or from a
- * hardware interrupt handler. Must be called with preemption disabled.
- * You may call it from a bottom half.
+ * hardware interrupt handler. You may call it from a bottom half.
*/
int smp_call_function_on(void (*func) (void *info), void *info, int nonatomic,
int wait, int cpu)
{
cpumask_t map = CPU_MASK_NONE;
+ preempt_disable();
cpu_set(cpu, map);
__smp_call_function_map(func, info, nonatomic, wait, map);
+ preempt_enable();
return 0;
}
EXPORT_SYMBOL(smp_call_function_on);
}
/*
- * Check which address space is addressed by the access
- * register in S390_lowcore.exc_access_id.
- * Returns 1 for user space and 0 for kernel space.
+ * Returns the address space associated with the fault.
+ * Returns 0 for kernel space, 1 for user space and
+ * 2 for code execution in user space with noexec=on.
*/
-static int __check_access_register(struct pt_regs *regs, int error_code)
-{
- int areg = S390_lowcore.exc_access_id;
-
- if (areg == 0)
- /* Access via access register 0 -> kernel address */
- return 0;
- save_access_regs(current->thread.acrs);
- if (regs && areg < NUM_ACRS && current->thread.acrs[areg] <= 1)
- /*
- * access register contains 0 -> kernel address,
- * access register contains 1 -> user space address
- */
- return current->thread.acrs[areg];
-
- /* Something unhealthy was done with the access registers... */
- die("page fault via unknown access register", regs, error_code);
- do_exit(SIGKILL);
- return 0;
-}
-
-/*
- * Check which address space the address belongs to.
- * May return 1 or 2 for user space and 0 for kernel space.
- * Returns 2 for user space in primary addressing mode with
- * CONFIG_S390_EXEC_PROTECT on and kernel parameter noexec=on.
- */
-static inline int check_user_space(struct pt_regs *regs, int error_code)
+static inline int check_space(struct task_struct *tsk)
{
/*
- * The lowest two bits of S390_lowcore.trans_exc_code indicate
- * which paging table was used:
- * 0: Primary Segment Table Descriptor
- * 1: STD determined via access register
- * 2: Secondary Segment Table Descriptor
- * 3: Home Segment Table Descriptor
+ * The lowest two bits of S390_lowcore.trans_exc_code
+ * indicate which paging table was used.
*/
- int descriptor = S390_lowcore.trans_exc_code & 3;
- if (unlikely(descriptor == 1))
- return __check_access_register(regs, error_code);
- if (descriptor == 2)
- return current->thread.mm_segment.ar4;
- return ((descriptor != 0) ^ (switch_amode)) << s390_noexec;
+ int desc = S390_lowcore.trans_exc_code & 3;
+
+ if (desc == 3) /* Home Segment Table Descriptor */
+ return switch_amode == 0;
+ if (desc == 2) /* Secondary Segment Table Descriptor */
+ return tsk->thread.mm_segment.ar4;
+#ifdef CONFIG_S390_SWITCH_AMODE
+ if (unlikely(desc == 1)) { /* STD determined via access register */
+ /* %a0 always indicates primary space. */
+ if (S390_lowcore.exc_access_id != 0) {
+ save_access_regs(tsk->thread.acrs);
+ /*
+ * An alet of 0 indicates primary space.
+ * An alet of 1 indicates secondary space.
+ * Any other alet values generate an
+ * alen-translation exception.
+ */
+ if (tsk->thread.acrs[S390_lowcore.exc_access_id])
+ return tsk->thread.mm_segment.ar4;
+ }
+ }
+#endif
+ /* Primary Segment Table Descriptor */
+ return switch_amode << s390_noexec;
}
/*
* 11 Page translation -> Not present (nullification)
* 3b Region third trans. -> Not present (nullification)
*/
-static inline void __kprobes
+static inline void
do_exception(struct pt_regs *regs, unsigned long error_code, int is_protection)
{
struct task_struct *tsk;
struct mm_struct *mm;
struct vm_area_struct * vma;
unsigned long address;
- int user_address;
const struct exception_table_entry *fixup;
- int si_code = SEGV_MAPERR;
+ int si_code;
+ int space;
tsk = current;
mm = tsk->mm;
NULL pointer write access in kernel mode. */
if (!(regs->psw.mask & PSW_MASK_PSTATE)) {
address = 0;
- user_address = 0;
+ space = 0;
goto no_context;
}
* the address
*/
address = S390_lowcore.trans_exc_code & __FAIL_ADDR_MASK;
- user_address = check_user_space(regs, error_code);
+ space = check_space(tsk);
/*
* Verify that the fault happened in user space, that
* we are not in an interrupt and that there is a
* user context.
*/
- if (user_address == 0 || in_atomic() || !mm)
- goto no_context;
+ if (unlikely(space == 0 || in_atomic() || !mm))
+ goto no_context;
/*
* When we get here, the fault happened in the current
down_read(&mm->mmap_sem);
- vma = find_vma(mm, address);
- if (!vma)
- goto bad_area;
+ si_code = SEGV_MAPERR;
+ vma = find_vma(mm, address);
+ if (!vma)
+ goto bad_area;
#ifdef CONFIG_S390_EXEC_PROTECT
- if (unlikely((user_address == 2) && !(vma->vm_flags & VM_EXEC)))
+ if (unlikely((space == 2) && !(vma->vm_flags & VM_EXEC)))
if (!signal_return(mm, regs, address, error_code))
/*
* signal_return() has done an up_read(&mm->mmap_sem)
* The instruction that caused the program check will
* be repeated. Don't signal single step via SIGTRAP.
*/
- clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
+ clear_tsk_thread_flag(tsk, TIF_SINGLE_STEP);
return;
/*
* Oops. The kernel tried to access some bad page. We'll have to
* terminate things with extreme prejudice.
*/
- if (user_address == 0)
+ if (space == 0)
printk(KERN_ALERT "Unable to handle kernel pointer dereference"
" at virtual kernel address %p\n", (void *)address);
else
goto no_context;
}
-void do_protection_exception(struct pt_regs *regs, unsigned long error_code)
+void __kprobes do_protection_exception(struct pt_regs *regs,
+ unsigned long error_code)
{
regs->psw.addr -= (error_code >> 16);
do_exception(regs, 4, 1);
}
-void do_dat_exception(struct pt_regs *regs, unsigned long error_code)
+void __kprobes do_dat_exception(struct pt_regs *regs, unsigned long error_code)
{
do_exception(regs, error_code & 0xff, 0);
}
cqr->device = device;
cqr->retries = 255;
cqr->expires = 10 * HZ;
+ clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
cqr->cpaddr->cmd_code = DASD_ECKD_CCW_SNSS;
cqr->cpaddr->count = SNSS_DATA_SIZE;
return PTR_ERR(request);
request->op = TO_NOP;
/* setup ccws */
- *device->modeset_byte = (mt_count == 0) ? 0x00 : 0x08;
+ if (mt_count == 0)
+ *device->modeset_byte &= ~0x08;
+ else
+ *device->modeset_byte |= 0x08;
tape_ccw_cc(request->cpaddr, MODE_SET_DB, 1, device->modeset_byte);
tape_ccw_end(request->cpaddr + 1, NOP, 0, NULL);
/* execute it */
ret = stsch(sch->schid, &sch->schib);
if (ret || !sch->schib.pmcw.dnv)
return -ENODEV;
- if (!sch->schib.pmcw.ena || sch->schib.scsw.actl == 0)
- /* Not operational or no activity -> done. */
+ if (!sch->schib.pmcw.ena)
+ /* Not operational -> done. */
return 0;
/* Stage 1: cancel io. */
if (!(sch->schib.scsw.actl & SCSW_ACTL_HALT_PEND) &&
struct ccw_device *cdev;
struct subchannel *sch;
int ret;
+ unsigned long flags;
priv = container_of(work, struct ccw_device_private, kick_work);
cdev = priv->cdev;
+ spin_lock_irqsave(cdev->ccwlock, flags);
sch = to_subchannel(cdev->dev.parent);
- ret = (sch->driver && sch->driver->notify) ?
- sch->driver->notify(&sch->dev, CIO_OPER) : 0;
- if (!ret)
- /* Driver doesn't want device back. */
- ccw_device_do_unreg_rereg(work);
- else {
+ if (sch->driver && sch->driver->notify) {
+ spin_unlock_irqrestore(cdev->ccwlock, flags);
+ ret = sch->driver->notify(&sch->dev, CIO_OPER);
+ spin_lock_irqsave(cdev->ccwlock, flags);
+ } else
+ ret = 0;
+ if (ret) {
/* Reenable channel measurements, if needed. */
+ spin_unlock_irqrestore(cdev->ccwlock, flags);
cmf_reenable(cdev);
+ spin_lock_irqsave(cdev->ccwlock, flags);
wake_up(&cdev->private->wait_q);
}
+ spin_unlock_irqrestore(cdev->ccwlock, flags);
+ if (!ret)
+ /* Driver doesn't want device back. */
+ ccw_device_do_unreg_rereg(work);
}
/*
struct ccw_device *cdev;
struct subchannel *sch;
int ret;
+ unsigned long flags;
priv = container_of(work, struct ccw_device_private, kick_work);
cdev = priv->cdev;
+ spin_lock_irqsave(cdev->ccwlock, flags);
sch = to_subchannel(cdev->dev.parent);
/* Extra sanity. */
if (sch->lpm)
- return;
- ret = (sch->driver && sch->driver->notify) ?
- sch->driver->notify(&sch->dev, CIO_NO_PATH) : 0;
+ goto out_unlock;
+ if (sch->driver && sch->driver->notify) {
+ spin_unlock_irqrestore(cdev->ccwlock, flags);
+ ret = sch->driver->notify(&sch->dev, CIO_NO_PATH);
+ spin_lock_irqsave(cdev->ccwlock, flags);
+ } else
+ ret = 0;
if (!ret) {
if (get_device(&sch->dev)) {
/* Driver doesn't want to keep device. */
cdev->private->state = DEV_STATE_DISCONNECTED;
wake_up(&cdev->private->wait_q);
}
+out_unlock:
+ spin_unlock_irqrestore(cdev->ccwlock, flags);
}
void
default:
/* Reset oper notify indication after verify error. */
cdev->private->flags.donotify = 0;
- PREPARE_WORK(&cdev->private->kick_work,
- ccw_device_nopath_notify);
- queue_work(ccw_device_notify_work, &cdev->private->kick_work);
- ccw_device_done(cdev, DEV_STATE_NOT_OPER);
+ if (cdev->online) {
+ PREPARE_WORK(&cdev->private->kick_work,
+ ccw_device_nopath_notify);
+ queue_work(ccw_device_notify_work,
+ &cdev->private->kick_work);
+ } else
+ ccw_device_done(cdev, DEV_STATE_NOT_OPER);
break;
}
}
ccw_device_online_notoper(struct ccw_device *cdev, enum dev_event dev_event)
{
struct subchannel *sch;
+ int ret;
sch = to_subchannel(cdev->dev.parent);
- if (sch->driver->notify &&
- sch->driver->notify(&sch->dev, sch->lpm ? CIO_GONE : CIO_NO_PATH)) {
- ccw_device_set_timeout(cdev, 0);
- cdev->private->flags.fake_irb = 0;
- cdev->private->state = DEV_STATE_DISCONNECTED;
- wake_up(&cdev->private->wait_q);
- return;
+ if (sch->driver->notify) {
+ spin_unlock_irq(cdev->ccwlock);
+ ret = sch->driver->notify(&sch->dev,
+ sch->lpm ? CIO_GONE : CIO_NO_PATH);
+ spin_lock_irq(cdev->ccwlock);
+ } else
+ ret = 0;
+ if (ret) {
+ ccw_device_set_timeout(cdev, 0);
+ cdev->private->flags.fake_irb = 0;
+ cdev->private->state = DEV_STATE_DISCONNECTED;
+ wake_up(&cdev->private->wait_q);
+ return;
}
cdev->private->state = DEV_STATE_NOT_OPER;
cio_disable_subchannel(sch);
sch = to_subchannel(cdev->dev.parent);
ccw_device_set_timeout(cdev, 0);
+ /* Start delayed path verification. */
+ ccw_device_online_verify(cdev, 0);
/* OK, i/o is dead now. Call interrupt handler. */
- cdev->private->state = DEV_STATE_ONLINE;
if (cdev->handler)
cdev->handler(cdev, cdev->private->intparm,
ERR_PTR(-EIO));
- if (!sch->lpm) {
- PREPARE_WORK(&cdev->private->kick_work,
- ccw_device_nopath_notify);
- queue_work(ccw_device_notify_work, &cdev->private->kick_work);
- } else if (cdev->private->flags.doverify)
- /* Start delayed path verification. */
- ccw_device_online_verify(cdev, 0);
}
static void
ccw_device_set_timeout(cdev, 3*HZ);
return;
}
- if (ret == -ENODEV) {
- struct subchannel *sch;
-
- sch = to_subchannel(cdev->dev.parent);
- if (!sch->lpm) {
- PREPARE_WORK(&cdev->private->kick_work,
- ccw_device_nopath_notify);
- queue_work(ccw_device_notify_work,
- &cdev->private->kick_work);
- } else
- dev_fsm_event(cdev, DEV_EVENT_NOTOPER);
- return;
- }
- //FIXME: Can we get here?
- cdev->private->state = DEV_STATE_ONLINE;
+ /* Start delayed path verification. */
+ ccw_device_online_verify(cdev, 0);
if (cdev->handler)
cdev->handler(cdev, cdev->private->intparm,
ERR_PTR(-EIO));
cdev->private->state = DEV_STATE_TIMEOUT_KILL;
return;
}
- if (ret == -ENODEV) {
- if (!sch->lpm) {
- PREPARE_WORK(&cdev->private->kick_work,
- ccw_device_nopath_notify);
- queue_work(ccw_device_notify_work,
- &cdev->private->kick_work);
- } else
- dev_fsm_event(cdev, DEV_EVENT_NOTOPER);
- return;
- }
+ /* Start delayed path verification. */
+ ccw_device_online_verify(cdev, 0);
if (cdev->handler)
cdev->handler(cdev, cdev->private->intparm,
ERR_PTR(-EIO));
- if (!sch->lpm) {
- PREPARE_WORK(&cdev->private->kick_work,
- ccw_device_nopath_notify);
- queue_work(ccw_device_notify_work, &cdev->private->kick_work);
- } else
- /* Start delayed path verification. */
- ccw_device_online_verify(cdev, 0);
}
static void
* void check_bugs(void);
*/
-static void __init check_bugs(void)
+static inline void check_bugs(void)
{
/* s390 has no bugs ... */
}
extern u32 ipl_flags;
extern u16 ipl_devno;
+extern u32 dump_prefix_page;
extern void do_reipl(void);
extern void ipl_save_parameters(void);
git.openpandora.org / pandora-kernel.git / commitdiff