From 2688905e6a9b3647bf7b452cb0ff2bdb166bd8fe Mon Sep 17 00:00:00 2001 From: Martin Schwidefsky Date: Wed, 7 May 2008 09:22:52 +0200 Subject: [PATCH 1/8] [S390] s390: Optimize user and work TIF check On return from syscall or interrupt, we have to check if we return to userspace (likely) and if there is work todo (less likely) to decide if we handle the work. We can optimize this check: we first check for the less likely work case and then check for userspace. This patch is also a preparation for an additional patch, that fixes a bug in KVM dealing with cpu bound guests. Signed-off-by: Martin Schwidefsky --- arch/s390/kernel/entry.S | 29 ++++++++++++++--------------- arch/s390/kernel/entry64.S | 29 ++++++++++++++--------------- 2 files changed, 28 insertions(+), 30 deletions(-) diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S index bdbb3bcd78a5..708cf9cf9a35 100644 --- a/arch/s390/kernel/entry.S +++ b/arch/s390/kernel/entry.S @@ -279,8 +279,6 @@ sysc_do_restart: st %r2,SP_R2(%r15) # store return value (change R2 on stack) sysc_return: - tm SP_PSW+1(%r15),0x01 # returning to user ? - 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: @@ -312,6 +310,8 @@ sysc_work_loop: # One of the work bits is on. Find out which one. # sysc_work: + tm SP_PSW+1(%r15),0x01 # returning to user ? + bno BASED(sysc_restore) tm __TI_flags+3(%r9),_TIF_MCCK_PENDING bo BASED(sysc_mcck_pending) tm __TI_flags+3(%r9),_TIF_NEED_RESCHED @@ -602,12 +602,6 @@ io_no_vtime: la %r2,SP_PTREGS(%r15) # address of register-save area basr %r14,%r1 # branch to standard irq handler 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_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: @@ -629,10 +623,18 @@ io_restore_trace_psw: .long 0, io_restore_trace + 0x80000000 #endif -#ifdef CONFIG_PREEMPT -io_preempt: +# +# switch to kernel stack, then check the TIF bits +# +io_work: + tm SP_PSW+1(%r15),0x01 # returning to user ? +#ifndef CONFIG_PREEMPT + bno BASED(io_restore) # no-> skip resched & signal +#else + bnz BASED(io_work_user) # no -> check for preemptive scheduling + # check for preemptive scheduling icm %r0,15,__TI_precount(%r9) - bnz BASED(io_restore) + bnz BASED(io_restore) # preemption disabled l %r1,SP_R15(%r15) s %r1,BASED(.Lc_spsize) mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) @@ -646,10 +648,7 @@ io_resume_loop: br %r1 # call schedule #endif -# -# switch to kernel stack, then check the TIF bits -# -io_work: +io_work_user: l %r1,__LC_KERNEL_STACK s %r1,BASED(.Lc_spsize) mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S index 5a4a7bcd2bba..a57909d63149 100644 --- a/arch/s390/kernel/entry64.S +++ b/arch/s390/kernel/entry64.S @@ -271,8 +271,6 @@ sysc_noemu: stg %r2,SP_R2(%r15) # store return value (change R2 on stack) sysc_return: - tm SP_PSW+1(%r15),0x01 # returning to user ? - jno sysc_restore tm __TI_flags+7(%r9),_TIF_WORK_SVC jnz sysc_work # there is work to do (signals etc.) sysc_restore: @@ -304,6 +302,8 @@ sysc_work_loop: # One of the work bits is on. Find out which one. # sysc_work: + tm SP_PSW+1(%r15),0x01 # returning to user ? + jno sysc_restore tm __TI_flags+7(%r9),_TIF_MCCK_PENDING jo sysc_mcck_pending tm __TI_flags+7(%r9),_TIF_NEED_RESCHED @@ -585,12 +585,6 @@ io_no_vtime: la %r2,SP_PTREGS(%r15) # address of register-save area brasl %r14,do_IRQ # call standard irq handler 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_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: @@ -612,10 +606,18 @@ io_restore_trace_psw: .quad 0, io_restore_trace #endif -#ifdef CONFIG_PREEMPT -io_preempt: +# +# switch to kernel stack, then check TIF bits +# +io_work: + tm SP_PSW+1(%r15),0x01 # returning to user ? +#ifndef CONFIG_PREEMPT + jno io_restore # no-> skip resched & signal +#else + jnz io_work_user # yes -> do resched & signal + # check for preemptive scheduling icm %r0,15,__TI_precount(%r9) - jnz io_restore + jnz io_restore # preemption is disabled # switch to kernel stack lg %r1,SP_R15(%r15) aghi %r1,-SP_SIZE @@ -629,10 +631,7 @@ io_resume_loop: jg preempt_schedule_irq #endif -# -# switch to kernel stack, then check TIF bits -# -io_work: +io_work_user: lg %r1,__LC_KERNEL_STACK aghi %r1,-SP_SIZE mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) From 0eaeafa10f3b2bd027e95859a6785d4c7fcc174c Mon Sep 17 00:00:00 2001 From: Christian Borntraeger Date: Wed, 7 May 2008 09:22:53 +0200 Subject: [PATCH 2/8] [S390] s390-kvm: leave sie context on work. Removes preemption requirement From: Martin Schwidefsky This patch fixes a bug with cpu bound guest on kvm-s390. Sometimes it was impossible to deliver a signal to a spinning guest. We used preemption as a circumvention. The preemption notifiers called vcpu_load, which checked for pending signals and triggered a host intercept. But even with preemption, a sigkill was not delivered immediately. This patch changes the low level host interrupt handler to check for the SIE instruction, if TIF_WORK is set. In that case we change the instruction pointer of the return PSW to rerun the vcpu_run loop. The kvm code sees an intercept reason 0 if that happens. This patch adds accounting for these types of intercept as well. The advantages: - works with and without preemption - signals are delivered immediately - much better host latencies without preemption Acked-by: Carsten Otte Signed-off-by: Christian Borntraeger Signed-off-by: Martin Schwidefsky --- arch/s390/kernel/entry64.S | 30 +++++++++++++++++++++++++++++- arch/s390/kvm/Kconfig | 1 - arch/s390/kvm/intercept.c | 3 +++ arch/s390/kvm/kvm-s390.c | 5 +---- include/asm-s390/kvm_host.h | 1 + 5 files changed, 34 insertions(+), 6 deletions(-) diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S index a57909d63149..fee10177dbfc 100644 --- a/arch/s390/kernel/entry64.S +++ b/arch/s390/kernel/entry64.S @@ -607,14 +607,37 @@ io_restore_trace_psw: #endif # -# switch to kernel stack, then check TIF bits +# There is work todo, we need to check if we return to userspace, then +# check, if we are in SIE, if yes leave it # io_work: tm SP_PSW+1(%r15),0x01 # returning to user ? #ifndef CONFIG_PREEMPT +#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) + jnz io_work_user # yes -> no need to check for SIE + la %r1, BASED(sie_opcode) # we return to kernel here + lg %r2, SP_PSW+8(%r15) + clc 0(2,%r1), 0(%r2) # is current instruction = SIE? + jne io_restore # no-> return to kernel + lg %r1, SP_PSW+8(%r15) # yes-> add 4 bytes to leave SIE + aghi %r1, 4 + stg %r1, SP_PSW+8(%r15) + j io_restore # return to kernel +#else jno io_restore # no-> skip resched & signal +#endif #else jnz io_work_user # yes -> do resched & signal +#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) + la %r1, BASED(sie_opcode) + lg %r2, SP_PSW+8(%r15) + clc 0(2,%r1), 0(%r2) # is current instruction = SIE? + jne 0f # no -> leave PSW alone + lg %r1, SP_PSW+8(%r15) # yes-> add 4 bytes to leave SIE + aghi %r1, 4 + stg %r1, SP_PSW+8(%r15) +0: +#endif # check for preemptive scheduling icm %r0,15,__TI_precount(%r9) jnz io_restore # preemption is disabled @@ -652,6 +675,11 @@ io_work_loop: j io_restore io_work_done: +#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) +sie_opcode: + .long 0xb2140000 +#endif + # # _TIF_MCCK_PENDING is set, call handler # diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig index 1761b74d639b..e051cad1f1e0 100644 --- a/arch/s390/kvm/Kconfig +++ b/arch/s390/kvm/Kconfig @@ -22,7 +22,6 @@ config KVM select PREEMPT_NOTIFIERS select ANON_INODES select S390_SWITCH_AMODE - select PREEMPT ---help--- Support hosting paravirtualized guest machines using the SIE virtualization capability on the mainframe. This should work diff --git a/arch/s390/kvm/intercept.c b/arch/s390/kvm/intercept.c index 349581a26103..47a0b642174c 100644 --- a/arch/s390/kvm/intercept.c +++ b/arch/s390/kvm/intercept.c @@ -105,6 +105,9 @@ static intercept_handler_t instruction_handlers[256] = { static int handle_noop(struct kvm_vcpu *vcpu) { switch (vcpu->arch.sie_block->icptcode) { + case 0x0: + vcpu->stat.exit_null++; + break; case 0x10: vcpu->stat.exit_external_request++; break; diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 98d1e73e01f1..0ac36a649eba 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -31,6 +31,7 @@ struct kvm_stats_debugfs_item debugfs_entries[] = { { "userspace_handled", VCPU_STAT(exit_userspace) }, + { "exit_null", VCPU_STAT(exit_null) }, { "exit_validity", VCPU_STAT(exit_validity) }, { "exit_stop_request", VCPU_STAT(exit_stop_request) }, { "exit_external_request", VCPU_STAT(exit_external_request) }, @@ -221,10 +222,6 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu) vcpu->arch.guest_fpregs.fpc &= FPC_VALID_MASK; restore_fp_regs(&vcpu->arch.guest_fpregs); restore_access_regs(vcpu->arch.guest_acrs); - - if (signal_pending(current)) - atomic_set_mask(CPUSTAT_STOP_INT, - &vcpu->arch.sie_block->cpuflags); } void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu) diff --git a/include/asm-s390/kvm_host.h b/include/asm-s390/kvm_host.h index f8204a4f2e02..18cbd8a39796 100644 --- a/include/asm-s390/kvm_host.h +++ b/include/asm-s390/kvm_host.h @@ -104,6 +104,7 @@ struct sie_block { struct kvm_vcpu_stat { u32 exit_userspace; + u32 exit_null; u32 exit_external_request; u32 exit_external_interrupt; u32 exit_stop_request; From 139b83dd57248a3c8fcfb256e562311ad61478e9 Mon Sep 17 00:00:00 2001 From: Michael Ernst Date: Wed, 7 May 2008 09:22:54 +0200 Subject: [PATCH 3/8] [S390] cio: Remove cio_msg kernel parameter. The only sporadically used CIO_DEBUG messages are replaced by ordinary CIO_MSG_EVENT messages. The CIO_MSG_EVENT messages debug levels are consolidated. Signed-off-by: Michael Ernst Signed-off-by: Martin Schwidefsky --- Documentation/s390/CommonIO | 11 --------- drivers/s390/cio/blacklist.c | 1 - drivers/s390/cio/cio.c | 39 ++++++++---------------------- drivers/s390/cio/cio.h | 2 -- drivers/s390/cio/cio_debug.h | 6 ----- drivers/s390/cio/css.c | 4 ++-- drivers/s390/cio/device.c | 25 ++++++++++--------- drivers/s390/cio/device_fsm.c | 44 +++++++++++++++++----------------- drivers/s390/cio/device_id.c | 4 ++-- drivers/s390/cio/device_pgid.c | 12 +++++----- 10 files changed, 54 insertions(+), 94 deletions(-) diff --git a/Documentation/s390/CommonIO b/Documentation/s390/CommonIO index 8fbc0a852870..bf0baa19ec24 100644 --- a/Documentation/s390/CommonIO +++ b/Documentation/s390/CommonIO @@ -8,17 +8,6 @@ Command line parameters Enable logging of debug information in case of ccw device timeouts. - -* cio_msg = yes | no - - Determines whether information on found devices and sensed device - characteristics should be shown during startup or when new devices are - found, i. e. messages of the types "Detected device 0.0.4711 on subchannel - 0.0.0042" and "SenseID: Device 0.0.4711 reports: ...". - - Default is off. - - * cio_ignore = {all} | { | } | {! | !} diff --git a/drivers/s390/cio/blacklist.c b/drivers/s390/cio/blacklist.c index 40ef948fcb3a..084447618999 100644 --- a/drivers/s390/cio/blacklist.c +++ b/drivers/s390/cio/blacklist.c @@ -199,7 +199,6 @@ blacklist_parse_parameters (char *str, range_action action) static int __init blacklist_setup (char *str) { - CIO_MSG_EVENT(6, "Reading blacklist parameters\n"); return blacklist_parse_parameters (str, add); } diff --git a/drivers/s390/cio/cio.c b/drivers/s390/cio/cio.c index 08a578161306..82c6a2d45128 100644 --- a/drivers/s390/cio/cio.c +++ b/drivers/s390/cio/cio.c @@ -39,23 +39,6 @@ debug_info_t *cio_debug_msg_id; debug_info_t *cio_debug_trace_id; debug_info_t *cio_debug_crw_id; -int cio_show_msg; - -static int __init -cio_setup (char *parm) -{ - if (!strcmp (parm, "yes")) - cio_show_msg = 1; - else if (!strcmp (parm, "no")) - cio_show_msg = 0; - else - printk(KERN_ERR "cio: cio_setup: " - "invalid cio_msg parameter '%s'", parm); - return 1; -} - -__setup ("cio_msg=", cio_setup); - /* * Function: cio_debug_init * Initializes three debug logs for common I/O: @@ -166,7 +149,7 @@ cio_start_handle_notoper(struct subchannel *sch, __u8 lpm) stsch (sch->schid, &sch->schib); - CIO_MSG_EVENT(0, "cio_start: 'not oper' status for " + CIO_MSG_EVENT(2, "cio_start: 'not oper' status for " "subchannel 0.%x.%04x!\n", sch->schid.ssid, sch->schid.sch_no); sprintf(dbf_text, "no%s", sch->dev.bus_id); @@ -567,10 +550,9 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid) * ... just being curious we check for non I/O subchannels */ if (sch->st != 0) { - CIO_DEBUG(KERN_INFO, 0, - "Subchannel 0.%x.%04x reports " - "non-I/O subchannel type %04X\n", - sch->schid.ssid, sch->schid.sch_no, sch->st); + CIO_MSG_EVENT(4, "Subchannel 0.%x.%04x reports " + "non-I/O subchannel type %04X\n", + sch->schid.ssid, sch->schid.sch_no, sch->st); /* We stop here for non-io subchannels. */ err = sch->st; goto out; @@ -588,7 +570,7 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid) * This device must not be known to Linux. So we simply * say that there is no device and return ENODEV. */ - CIO_MSG_EVENT(4, "Blacklisted device detected " + CIO_MSG_EVENT(6, "Blacklisted device detected " "at devno %04X, subchannel set %x\n", sch->schib.pmcw.dev, sch->schid.ssid); err = -ENODEV; @@ -601,12 +583,11 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid) sch->lpm = sch->schib.pmcw.pam & sch->opm; sch->isc = 3; - CIO_DEBUG(KERN_INFO, 0, - "Detected device %04x on subchannel 0.%x.%04X" - " - PIM = %02X, PAM = %02X, POM = %02X\n", - sch->schib.pmcw.dev, sch->schid.ssid, - sch->schid.sch_no, sch->schib.pmcw.pim, - sch->schib.pmcw.pam, sch->schib.pmcw.pom); + CIO_MSG_EVENT(6, "Detected device %04x on subchannel 0.%x.%04X " + "- PIM = %02X, PAM = %02X, POM = %02X\n", + sch->schib.pmcw.dev, sch->schid.ssid, + sch->schid.sch_no, sch->schib.pmcw.pim, + sch->schib.pmcw.pam, sch->schib.pmcw.pom); /* * We now have to initially ... diff --git a/drivers/s390/cio/cio.h b/drivers/s390/cio/cio.h index 3c75412904dc..6e933aebe013 100644 --- a/drivers/s390/cio/cio.h +++ b/drivers/s390/cio/cio.h @@ -118,6 +118,4 @@ extern void *cio_get_console_priv(void); #define cio_get_console_priv() NULL #endif -extern int cio_show_msg; - #endif diff --git a/drivers/s390/cio/cio_debug.h b/drivers/s390/cio/cio_debug.h index d7429ef6c666..e64e8278c42e 100644 --- a/drivers/s390/cio/cio_debug.h +++ b/drivers/s390/cio/cio_debug.h @@ -31,10 +31,4 @@ static inline void CIO_HEX_EVENT(int level, void *data, int length) } } -#define CIO_DEBUG(printk_level, event_level, msg...) do { \ - if (cio_show_msg) \ - printk(printk_level "cio: " msg); \ - CIO_MSG_EVENT(event_level, msg); \ - } while (0) - #endif diff --git a/drivers/s390/cio/css.c b/drivers/s390/cio/css.c index 595e327d2f76..a76956512b2d 100644 --- a/drivers/s390/cio/css.c +++ b/drivers/s390/cio/css.c @@ -570,7 +570,7 @@ static void reprobe_all(struct work_struct *unused) { int ret; - CIO_MSG_EVENT(2, "reprobe start\n"); + CIO_MSG_EVENT(4, "reprobe start\n"); need_reprobe = 0; /* Make sure initial subchannel scan is done. */ @@ -578,7 +578,7 @@ static void reprobe_all(struct work_struct *unused) atomic_read(&ccw_device_init_count) == 0); ret = for_each_subchannel_staged(NULL, reprobe_subchannel, NULL); - CIO_MSG_EVENT(2, "reprobe done (rc=%d, need_reprobe=%d)\n", ret, + CIO_MSG_EVENT(4, "reprobe done (rc=%d, need_reprobe=%d)\n", ret, need_reprobe); } diff --git a/drivers/s390/cio/device.c b/drivers/s390/cio/device.c index abfd601d237a..e22813db74a2 100644 --- a/drivers/s390/cio/device.c +++ b/drivers/s390/cio/device.c @@ -341,7 +341,7 @@ ccw_device_remove_disconnected(struct ccw_device *cdev) rc = device_schedule_callback(&cdev->dev, ccw_device_remove_orphan_cb); if (rc) - CIO_MSG_EVENT(2, "Couldn't unregister orphan " + CIO_MSG_EVENT(0, "Couldn't unregister orphan " "0.%x.%04x\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -351,7 +351,7 @@ ccw_device_remove_disconnected(struct ccw_device *cdev) rc = device_schedule_callback(cdev->dev.parent, ccw_device_remove_sch_cb); if (rc) - CIO_MSG_EVENT(2, "Couldn't unregister disconnected device " + CIO_MSG_EVENT(0, "Couldn't unregister disconnected device " "0.%x.%04x\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -397,7 +397,7 @@ int ccw_device_set_offline(struct ccw_device *cdev) if (ret == 0) wait_event(cdev->private->wait_q, dev_fsm_final_state(cdev)); else { - CIO_MSG_EVENT(2, "ccw_device_offline returned %d, " + CIO_MSG_EVENT(0, "ccw_device_offline returned %d, " "device 0.%x.%04x\n", ret, cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -433,7 +433,7 @@ int ccw_device_set_online(struct ccw_device *cdev) if (ret == 0) wait_event(cdev->private->wait_q, dev_fsm_final_state(cdev)); else { - CIO_MSG_EVENT(2, "ccw_device_online returned %d, " + CIO_MSG_EVENT(0, "ccw_device_online returned %d, " "device 0.%x.%04x\n", ret, cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -451,7 +451,7 @@ int ccw_device_set_online(struct ccw_device *cdev) if (ret == 0) wait_event(cdev->private->wait_q, dev_fsm_final_state(cdev)); else - CIO_MSG_EVENT(2, "ccw_device_offline returned %d, " + CIO_MSG_EVENT(0, "ccw_device_offline returned %d, " "device 0.%x.%04x\n", ret, cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -803,7 +803,7 @@ static void sch_attach_disconnected_device(struct subchannel *sch, other_sch = to_subchannel(get_device(cdev->dev.parent)); ret = device_move(&cdev->dev, &sch->dev); if (ret) { - CIO_MSG_EVENT(2, "Moving disconnected device 0.%x.%04x failed " + CIO_MSG_EVENT(0, "Moving disconnected device 0.%x.%04x failed " "(ret=%d)!\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno, ret); put_device(&other_sch->dev); @@ -933,7 +933,7 @@ io_subchannel_register(struct work_struct *work) ret = device_reprobe(&cdev->dev); if (ret) /* We can't do much here. */ - CIO_MSG_EVENT(2, "device_reprobe() returned" + CIO_MSG_EVENT(0, "device_reprobe() returned" " %d for 0.%x.%04x\n", ret, cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -1086,7 +1086,7 @@ static void ccw_device_move_to_sch(struct work_struct *work) rc = device_move(&cdev->dev, &sch->dev); mutex_unlock(&sch->reg_mutex); if (rc) { - CIO_MSG_EVENT(2, "Moving device 0.%x.%04x to subchannel " + CIO_MSG_EVENT(0, "Moving device 0.%x.%04x to subchannel " "0.%x.%04x failed (ret=%d)!\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno, sch->schid.ssid, @@ -1446,8 +1446,7 @@ ccw_device_remove (struct device *dev) wait_event(cdev->private->wait_q, dev_fsm_final_state(cdev)); else - //FIXME: we can't fail! - CIO_MSG_EVENT(2, "ccw_device_offline returned %d, " + CIO_MSG_EVENT(0, "ccw_device_offline returned %d, " "device 0.%x.%04x\n", ret, cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -1524,7 +1523,7 @@ static int recovery_check(struct device *dev, void *data) spin_lock_irq(cdev->ccwlock); switch (cdev->private->state) { case DEV_STATE_DISCONNECTED: - CIO_MSG_EVENT(3, "recovery: trigger 0.%x.%04x\n", + CIO_MSG_EVENT(4, "recovery: trigger 0.%x.%04x\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno); dev_fsm_event(cdev, DEV_EVENT_VERIFY); @@ -1554,7 +1553,7 @@ static void recovery_work_func(struct work_struct *unused) } spin_unlock_irq(&recovery_lock); } else - CIO_MSG_EVENT(2, "recovery: end\n"); + CIO_MSG_EVENT(4, "recovery: end\n"); } static DECLARE_WORK(recovery_work, recovery_work_func); @@ -1572,7 +1571,7 @@ void ccw_device_schedule_recovery(void) { unsigned long flags; - CIO_MSG_EVENT(2, "recovery: schedule\n"); + CIO_MSG_EVENT(4, "recovery: schedule\n"); spin_lock_irqsave(&recovery_lock, flags); if (!timer_pending(&recovery_timer) || (recovery_phase != 0)) { recovery_phase = 0; diff --git a/drivers/s390/cio/device_fsm.c b/drivers/s390/cio/device_fsm.c index 99403b0a97a7..e268d5a77c12 100644 --- a/drivers/s390/cio/device_fsm.c +++ b/drivers/s390/cio/device_fsm.c @@ -322,10 +322,10 @@ ccw_device_recog_done(struct ccw_device *cdev, int state) same_dev = 0; /* Keep the compiler quiet... */ switch (state) { case DEV_STATE_NOT_OPER: - CIO_DEBUG(KERN_WARNING, 2, - "SenseID : unknown device %04x on subchannel " - "0.%x.%04x\n", cdev->private->dev_id.devno, - sch->schid.ssid, sch->schid.sch_no); + CIO_MSG_EVENT(2, "SenseID : unknown device %04x on " + "subchannel 0.%x.%04x\n", + cdev->private->dev_id.devno, + sch->schid.ssid, sch->schid.sch_no); break; case DEV_STATE_OFFLINE: if (cdev->private->state == DEV_STATE_DISCONNECTED_SENSE_ID) { @@ -348,20 +348,19 @@ ccw_device_recog_done(struct ccw_device *cdev, int state) return; } /* Issue device info message. */ - CIO_DEBUG(KERN_INFO, 2, - "SenseID : device 0.%x.%04x reports: " - "CU Type/Mod = %04X/%02X, Dev Type/Mod = " - "%04X/%02X\n", - cdev->private->dev_id.ssid, - cdev->private->dev_id.devno, - cdev->id.cu_type, cdev->id.cu_model, - cdev->id.dev_type, cdev->id.dev_model); + CIO_MSG_EVENT(4, "SenseID : device 0.%x.%04x reports: " + "CU Type/Mod = %04X/%02X, Dev Type/Mod = " + "%04X/%02X\n", + cdev->private->dev_id.ssid, + cdev->private->dev_id.devno, + cdev->id.cu_type, cdev->id.cu_model, + cdev->id.dev_type, cdev->id.dev_model); break; case DEV_STATE_BOXED: - CIO_DEBUG(KERN_WARNING, 2, - "SenseID : boxed device %04x on subchannel " - "0.%x.%04x\n", cdev->private->dev_id.devno, - sch->schid.ssid, sch->schid.sch_no); + CIO_MSG_EVENT(0, "SenseID : boxed device %04x on " + " subchannel 0.%x.%04x\n", + cdev->private->dev_id.devno, + sch->schid.ssid, sch->schid.sch_no); break; } cdev->private->state = state; @@ -443,9 +442,8 @@ ccw_device_done(struct ccw_device *cdev, int state) if (state == DEV_STATE_BOXED) - CIO_DEBUG(KERN_WARNING, 2, - "Boxed device %04x on subchannel %04x\n", - cdev->private->dev_id.devno, sch->schid.sch_no); + CIO_MSG_EVENT(0, "Boxed device %04x on subchannel %04x\n", + cdev->private->dev_id.devno, sch->schid.sch_no); if (cdev->private->flags.donotify) { cdev->private->flags.donotify = 0; @@ -900,7 +898,7 @@ ccw_device_w4sense(struct ccw_device *cdev, enum dev_event dev_event) /* Basic sense hasn't started. Try again. */ ccw_device_do_sense(cdev, irb); else { - CIO_MSG_EVENT(2, "Huh? 0.%x.%04x: unsolicited " + CIO_MSG_EVENT(0, "0.%x.%04x: unsolicited " "interrupt during w4sense...\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno); @@ -1169,8 +1167,10 @@ ccw_device_nop(struct ccw_device *cdev, enum dev_event dev_event) static void ccw_device_bug(struct ccw_device *cdev, enum dev_event dev_event) { - CIO_MSG_EVENT(0, "dev_jumptable[%i][%i] == NULL\n", - cdev->private->state, dev_event); + CIO_MSG_EVENT(0, "Internal state [%i][%i] not handled for device " + "0.%x.%04x\n", cdev->private->state, dev_event, + cdev->private->dev_id.ssid, + cdev->private->dev_id.devno); BUG(); } diff --git a/drivers/s390/cio/device_id.c b/drivers/s390/cio/device_id.c index dc4d87f77f6c..cba7020517ed 100644 --- a/drivers/s390/cio/device_id.c +++ b/drivers/s390/cio/device_id.c @@ -214,7 +214,7 @@ ccw_device_check_sense_id(struct ccw_device *cdev) * sense id information. So, for intervention required, * we use the "whack it until it talks" strategy... */ - CIO_MSG_EVENT(2, "SenseID : device %04x on Subchannel " + CIO_MSG_EVENT(0, "SenseID : device %04x on Subchannel " "0.%x.%04x reports cmd reject\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no); @@ -239,7 +239,7 @@ ccw_device_check_sense_id(struct ccw_device *cdev) lpm = to_io_private(sch)->orb.lpm; if ((lpm & sch->schib.pmcw.pim & sch->schib.pmcw.pam) != 0) - CIO_MSG_EVENT(2, "SenseID : path %02X for device %04x " + CIO_MSG_EVENT(4, "SenseID : path %02X for device %04x " "on subchannel 0.%x.%04x is " "'not operational'\n", lpm, cdev->private->dev_id.devno, diff --git a/drivers/s390/cio/device_pgid.c b/drivers/s390/cio/device_pgid.c index c52449a1f9fc..ba559053402e 100644 --- a/drivers/s390/cio/device_pgid.c +++ b/drivers/s390/cio/device_pgid.c @@ -79,7 +79,7 @@ __ccw_device_sense_pgid_start(struct ccw_device *cdev) /* ret is 0, -EBUSY, -EACCES or -ENODEV */ if (ret != -EACCES) return ret; - CIO_MSG_EVENT(2, "SNID - Device %04x on Subchannel " + CIO_MSG_EVENT(3, "SNID - Device %04x on Subchannel " "0.%x.%04x, lpm %02X, became 'not " "operational'\n", cdev->private->dev_id.devno, @@ -159,7 +159,7 @@ __ccw_device_check_sense_pgid(struct ccw_device *cdev) u8 lpm; lpm = to_io_private(sch)->orb.lpm; - CIO_MSG_EVENT(2, "SNID - Device %04x on Subchannel 0.%x.%04x," + CIO_MSG_EVENT(3, "SNID - Device %04x on Subchannel 0.%x.%04x," " lpm %02X, became 'not operational'\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no, lpm); @@ -275,7 +275,7 @@ __ccw_device_do_pgid(struct ccw_device *cdev, __u8 func) return ret; } /* PGID command failed on this path. */ - CIO_MSG_EVENT(2, "SPID - Device %04x on Subchannel " + CIO_MSG_EVENT(3, "SPID - Device %04x on Subchannel " "0.%x.%04x, lpm %02X, became 'not operational'\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no, cdev->private->imask); @@ -317,7 +317,7 @@ static int __ccw_device_do_nop(struct ccw_device *cdev) return ret; } /* nop command failed on this path. */ - CIO_MSG_EVENT(2, "NOP - Device %04x on Subchannel " + CIO_MSG_EVENT(3, "NOP - Device %04x on Subchannel " "0.%x.%04x, lpm %02X, became 'not operational'\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no, cdev->private->imask); @@ -362,7 +362,7 @@ __ccw_device_check_pgid(struct ccw_device *cdev) return -EAGAIN; } if (irb->scsw.cc == 3) { - CIO_MSG_EVENT(2, "SPID - Device %04x on Subchannel 0.%x.%04x," + CIO_MSG_EVENT(3, "SPID - Device %04x on Subchannel 0.%x.%04x," " lpm %02X, became 'not operational'\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no, cdev->private->imask); @@ -391,7 +391,7 @@ static int __ccw_device_check_nop(struct ccw_device *cdev) return -ETIME; } if (irb->scsw.cc == 3) { - CIO_MSG_EVENT(2, "NOP - Device %04x on Subchannel 0.%x.%04x," + CIO_MSG_EVENT(3, "NOP - Device %04x on Subchannel 0.%x.%04x," " lpm %02X, became 'not operational'\n", cdev->private->dev_id.devno, sch->schid.ssid, sch->schid.sch_no, cdev->private->imask); From 5b8909871b80a6cc2bd21aa5262c1424e3d26339 Mon Sep 17 00:00:00 2001 From: Michael Ernst Date: Wed, 7 May 2008 09:22:55 +0200 Subject: [PATCH 4/8] [S390] cio: Fix parsing mechanism for blacklisted devices. New format cssid.ssid.devno is now parsed correctly. Signed-off-by: Michael Ernst Signed-off-by: Martin Schwidefsky --- drivers/s390/cio/blacklist.c | 338 ++++++++++++++++++----------------- 1 file changed, 177 insertions(+), 161 deletions(-) diff --git a/drivers/s390/cio/blacklist.c b/drivers/s390/cio/blacklist.c index 084447618999..9c21b8f43f9b 100644 --- a/drivers/s390/cio/blacklist.c +++ b/drivers/s390/cio/blacklist.c @@ -19,6 +19,7 @@ #include #include +#include #include "blacklist.h" #include "cio.h" @@ -43,163 +44,169 @@ typedef enum {add, free} range_action; * Function: blacklist_range * (Un-)blacklist the devices from-to */ -static void -blacklist_range (range_action action, unsigned int from, unsigned int to, - unsigned int ssid) +static int blacklist_range(range_action action, unsigned int from_ssid, + unsigned int to_ssid, unsigned int from, + unsigned int to, int msgtrigger) { - if (!to) - to = from; - - if (from > to || to > __MAX_SUBCHANNEL || ssid > __MAX_SSID) { - printk (KERN_WARNING "cio: Invalid blacklist range " - "0.%x.%04x to 0.%x.%04x, skipping\n", - ssid, from, ssid, to); - return; + if ((from_ssid > to_ssid) || ((from_ssid == to_ssid) && (from > to))) { + if (msgtrigger) + printk(KERN_WARNING "cio: Invalid cio_ignore range " + "0.%x.%04x-0.%x.%04x\n", from_ssid, from, + to_ssid, to); + return 1; } - for (; from <= to; from++) { + + while ((from_ssid < to_ssid) || ((from_ssid == to_ssid) && + (from <= to))) { if (action == add) - set_bit (from, bl_dev[ssid]); + set_bit(from, bl_dev[from_ssid]); else - clear_bit (from, bl_dev[ssid]); + clear_bit(from, bl_dev[from_ssid]); + from++; + if (from > __MAX_SUBCHANNEL) { + from_ssid++; + from = 0; + } } -} -/* - * Function: blacklist_busid - * Get devno/busid from given string. - * Shamelessly grabbed from dasd_devmap.c. - */ -static int -blacklist_busid(char **str, int *id0, int *ssid, int *devno) -{ - int val, old_style; - char *sav; - - sav = *str; - - /* check for leading '0x' */ - old_style = 0; - if ((*str)[0] == '0' && (*str)[1] == 'x') { - *str += 2; - old_style = 1; - } - if (!isxdigit((*str)[0])) /* We require at least one hex digit */ - goto confused; - val = simple_strtoul(*str, str, 16); - if (old_style || (*str)[0] != '.') { - *id0 = *ssid = 0; - if (val < 0 || val > 0xffff) - goto confused; - *devno = val; - if ((*str)[0] != ',' && (*str)[0] != '-' && - (*str)[0] != '\n' && (*str)[0] != '\0') - goto confused; - return 0; - } - /* New style x.y.z busid */ - if (val < 0 || val > 0xff) - goto confused; - *id0 = val; - (*str)++; - if (!isxdigit((*str)[0])) /* We require at least one hex digit */ - goto confused; - val = simple_strtoul(*str, str, 16); - if (val < 0 || val > 0xff || (*str)++[0] != '.') - goto confused; - *ssid = val; - if (!isxdigit((*str)[0])) /* We require at least one hex digit */ - goto confused; - val = simple_strtoul(*str, str, 16); - if (val < 0 || val > 0xffff) - goto confused; - *devno = val; - if ((*str)[0] != ',' && (*str)[0] != '-' && - (*str)[0] != '\n' && (*str)[0] != '\0') - goto confused; return 0; -confused: - strsep(str, ",\n"); - printk(KERN_WARNING "cio: Invalid cio_ignore parameter '%s'\n", sav); - return 1; } -static int -blacklist_parse_parameters (char *str, range_action action) +static int pure_hex(char **cp, unsigned int *val, int min_digit, + int max_digit, int max_val) { - int from, to, from_id0, to_id0, from_ssid, to_ssid; + int diff; + unsigned int value; - while (*str != 0 && *str != '\n') { - range_action ra = action; - while(*str == ',') - str++; - if (*str == '!') { - ra = !action; - ++str; - } + diff = 0; + *val = 0; - /* - * Since we have to parse the proc commands and the - * kernel arguments we have to check four cases - */ - if (strncmp(str,"all,",4) == 0 || strcmp(str,"all") == 0 || - strncmp(str,"all\n",4) == 0 || strncmp(str,"all ",4) == 0) { - int j; + while (isxdigit(**cp) && (diff <= max_digit)) { - str += 3; - for (j=0; j <= __MAX_SSID; j++) - blacklist_range(ra, 0, __MAX_SUBCHANNEL, j); - } else { - int rc; - - rc = blacklist_busid(&str, &from_id0, - &from_ssid, &from); - if (rc) - continue; - to = from; - to_id0 = from_id0; - to_ssid = from_ssid; - if (*str == '-') { - str++; - rc = blacklist_busid(&str, &to_id0, - &to_ssid, &to); - if (rc) - continue; - } - if (*str == '-') { - printk(KERN_WARNING "cio: invalid cio_ignore " - "parameter '%s'\n", - strsep(&str, ",\n")); - continue; - } - if ((from_id0 != to_id0) || - (from_ssid != to_ssid)) { - printk(KERN_WARNING "cio: invalid cio_ignore " - "range %x.%x.%04x-%x.%x.%04x\n", - from_id0, from_ssid, from, - to_id0, to_ssid, to); - continue; - } - blacklist_range (ra, from, to, to_ssid); - } + if (isdigit(**cp)) + value = **cp - '0'; + else + value = tolower(**cp) - 'a' + 10; + *val = *val * 16 + value; + (*cp)++; + diff++; } - return 1; + + if ((diff < min_digit) || (diff > max_digit) || (*val > max_val)) + return 1; + + return 0; +} + +static int parse_busid(char *str, int *cssid, int *ssid, int *devno, + int msgtrigger) +{ + char *str_work; + int val, rc, ret; + + rc = 1; + + if (*str == '\0') + goto out; + + /* old style */ + str_work = str; + val = simple_strtoul(str, &str_work, 16); + + if (*str_work == '\0') { + if (val <= __MAX_SUBCHANNEL) { + *devno = val; + *ssid = 0; + *cssid = 0; + rc = 0; + } + goto out; + } + + /* new style */ + str_work = str; + ret = pure_hex(&str_work, cssid, 1, 2, __MAX_CSSID); + if (ret || (str_work[0] != '.')) + goto out; + str_work++; + ret = pure_hex(&str_work, ssid, 1, 1, __MAX_SSID); + if (ret || (str_work[0] != '.')) + goto out; + str_work++; + ret = pure_hex(&str_work, devno, 4, 4, __MAX_SUBCHANNEL); + if (ret || (str_work[0] != '\0')) + goto out; + + rc = 0; +out: + if (rc && msgtrigger) + printk(KERN_WARNING "cio: Invalid cio_ignore device '%s'\n", + str); + + return rc; +} + +static int blacklist_parse_parameters(char *str, range_action action, + int msgtrigger) +{ + int from_cssid, to_cssid, from_ssid, to_ssid, from, to; + int rc, totalrc; + char *parm; + range_action ra; + + totalrc = 0; + + while ((parm = strsep(&str, ","))) { + rc = 0; + ra = action; + if (*parm == '!') { + if (ra == add) + ra = free; + else + ra = add; + parm++; + } + if (strcmp(parm, "all") == 0) { + from_cssid = 0; + from_ssid = 0; + from = 0; + to_cssid = __MAX_CSSID; + to_ssid = __MAX_SSID; + to = __MAX_SUBCHANNEL; + } else { + rc = parse_busid(strsep(&parm, "-"), &from_cssid, + &from_ssid, &from, msgtrigger); + if (!rc) { + if (parm != NULL) + rc = parse_busid(parm, &to_cssid, + &to_ssid, &to, + msgtrigger); + else { + to_cssid = from_cssid; + to_ssid = from_ssid; + to = from; + } + } + } + if (!rc) { + rc = blacklist_range(ra, from_ssid, to_ssid, from, to, + msgtrigger); + if (rc) + totalrc = 1; + } else + totalrc = 1; + } + + return totalrc; } -/* Parsing the commandline for blacklist parameters, e.g. to blacklist - * bus ids 0.0.1234, 0.0.1235 and 0.0.1236, you could use any of: - * - cio_ignore=1234-1236 - * - cio_ignore=0x1234-0x1235,1236 - * - cio_ignore=0x1234,1235-1236 - * - cio_ignore=1236 cio_ignore=1234-0x1236 - * - cio_ignore=1234 cio_ignore=1236 cio_ignore=0x1235 - * - cio_ignore=0.0.1234-0.0.1236 - * - cio_ignore=0.0.1234,0x1235,1236 - * - ... - */ static int __init blacklist_setup (char *str) { - return blacklist_parse_parameters (str, add); + CIO_MSG_EVENT(6, "Reading blacklist parameters\n"); + if (blacklist_parse_parameters(str, add, 1)) + return 0; + return 1; } __setup ("cio_ignore=", blacklist_setup); @@ -223,27 +230,23 @@ is_blacklisted (int ssid, int devno) * Function: blacklist_parse_proc_parameters * parse the stuff which is piped to /proc/cio_ignore */ -static void -blacklist_parse_proc_parameters (char *buf) +static int blacklist_parse_proc_parameters(char *buf) { - if (strncmp (buf, "free ", 5) == 0) { - blacklist_parse_parameters (buf + 5, free); - } else if (strncmp (buf, "add ", 4) == 0) { - /* - * We don't need to check for known devices since - * css_probe_device will handle this correctly. - */ - blacklist_parse_parameters (buf + 4, add); - } else { - printk (KERN_WARNING "cio: cio_ignore: Parse error; \n" - KERN_WARNING "try using 'free all|," - ",...'\n" - KERN_WARNING "or 'add ," - ",...'\n"); - return; - } + int rc; + char *parm; + + parm = strsep(&buf, " "); + + if (strcmp("free", parm) == 0) + rc = blacklist_parse_parameters(buf, free, 0); + else if (strcmp("add", parm) == 0) + rc = blacklist_parse_parameters(buf, add, 0); + else + return 1; css_schedule_reprobe(); + + return rc; } /* Iterator struct for all devices. */ @@ -327,6 +330,8 @@ cio_ignore_write(struct file *file, const char __user *user_buf, size_t user_len, loff_t *offset) { char *buf; + size_t i; + ssize_t rc, ret; if (*offset) return -EINVAL; @@ -335,16 +340,27 @@ cio_ignore_write(struct file *file, const char __user *user_buf, buf = vmalloc (user_len + 1); /* maybe better use the stack? */ if (buf == NULL) return -ENOMEM; + memset(buf, 0, user_len + 1); + if (strncpy_from_user (buf, user_buf, user_len) < 0) { - vfree (buf); - return -EFAULT; + rc = -EFAULT; + goto out_free; } - buf[user_len] = '\0'; - blacklist_parse_proc_parameters (buf); + i = user_len - 1; + while ((i >= 0) && (isspace(buf[i]) || (buf[i] == 0))) { + buf[i] = '\0'; + i--; + } + ret = blacklist_parse_proc_parameters(buf); + if (ret) + rc = -EINVAL; + else + rc = user_len; +out_free: vfree (buf); - return user_len; + return rc; } static const struct seq_operations cio_ignore_proc_seq_ops = { From c6ca1850e78d60c299ceb4c240a04af9e2384f70 Mon Sep 17 00:00:00 2001 From: Martin Schwidefsky Date: Wed, 7 May 2008 09:22:56 +0200 Subject: [PATCH 5/8] [S390] s390mach compile warning Fix the following compile warning: drivers/s390/s390mach.c: In function 's390_collect_crw_info': drivers/s390/s390mach.c:77: warning: ignoring return value of 'down_interruptibl Signed-off-by: Martin Schwidefsky --- drivers/s390/s390mach.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/s390/s390mach.c b/drivers/s390/s390mach.c index 4d4b54277c43..5080f343ad74 100644 --- a/drivers/s390/s390mach.c +++ b/drivers/s390/s390mach.c @@ -48,10 +48,11 @@ s390_collect_crw_info(void *param) int ccode; struct semaphore *sem; unsigned int chain; + int ignore; sem = (struct semaphore *)param; repeat: - down_interruptible(sem); + ignore = down_interruptible(sem); chain = 0; while (1) { if (unlikely(chain > 1)) { From b499d76bfd78e900039155247e1c21bfdf807b7b Mon Sep 17 00:00:00 2001 From: Roland McGrath Date: Wed, 7 May 2008 09:22:57 +0200 Subject: [PATCH 6/8] [S390] compat ptrace cleanup This removes redundant arch code for generic ptrace requests already handled by ptrace_request and compat_ptrace_request. It simplifies things to just have the standard entry points, and use the generic compat_sys_ptrace. Signed-off-by: Roland McGrath Signed-off-by: Martin Schwidefsky --- arch/s390/kernel/compat_wrapper.S | 2 +- arch/s390/kernel/ptrace.c | 100 ++---------------------------- include/asm-s390/ptrace.h | 2 + 3 files changed, 9 insertions(+), 95 deletions(-) diff --git a/arch/s390/kernel/compat_wrapper.S b/arch/s390/kernel/compat_wrapper.S index 743d54f0b8db..d003a6e16afb 100644 --- a/arch/s390/kernel/compat_wrapper.S +++ b/arch/s390/kernel/compat_wrapper.S @@ -121,7 +121,7 @@ sys32_ptrace_wrapper: lgfr %r3,%r3 # long llgtr %r4,%r4 # long llgfr %r5,%r5 # long - jg sys_ptrace # branch to system call + jg compat_sys_ptrace # branch to system call .globl sys32_alarm_wrapper sys32_alarm_wrapper: diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c index 7f4270163744..35827b9bd4d1 100644 --- a/arch/s390/kernel/ptrace.c +++ b/arch/s390/kernel/ptrace.c @@ -292,8 +292,7 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data) return 0; } -static int -do_ptrace_normal(struct task_struct *child, long request, long addr, long data) +long arch_ptrace(struct task_struct *child, long request, long addr, long data) { ptrace_area parea; int copied, ret; @@ -529,35 +528,19 @@ poke_user_emu31(struct task_struct *child, addr_t addr, addr_t data) return 0; } -static int -do_ptrace_emu31(struct task_struct *child, long request, long addr, long data) +long compat_arch_ptrace(struct task_struct *child, compat_long_t request, + compat_ulong_t caddr, compat_ulong_t cdata) { - unsigned int tmp; /* 4 bytes !! */ + unsigned long addr = caddr; + unsigned long data = cdata; ptrace_area_emu31 parea; int copied, ret; switch (request) { - case PTRACE_PEEKTEXT: - case PTRACE_PEEKDATA: - /* read word at location addr. */ - copied = access_process_vm(child, addr, &tmp, sizeof(tmp), 0); - if (copied != sizeof(tmp)) - return -EIO; - return put_user(tmp, (unsigned int __force __user *) data); - case PTRACE_PEEKUSR: /* read the word at location addr in the USER area. */ return peek_user_emu31(child, addr, data); - case PTRACE_POKETEXT: - case PTRACE_POKEDATA: - /* write the word at location addr. */ - tmp = data; - copied = access_process_vm(child, addr, &tmp, sizeof(tmp), 1); - if (copied != sizeof(tmp)) - return -EIO; - return 0; - case PTRACE_POKEUSR: /* write the word at location addr in the USER area */ return poke_user_emu31(child, addr, data); @@ -587,82 +570,11 @@ do_ptrace_emu31(struct task_struct *child, long request, long addr, long data) copied += sizeof(unsigned int); } return 0; - case PTRACE_GETEVENTMSG: - return put_user((__u32) child->ptrace_message, - (unsigned int __force __user *) data); - case PTRACE_GETSIGINFO: - if (child->last_siginfo == NULL) - return -EINVAL; - return copy_siginfo_to_user32((compat_siginfo_t - __force __user *) data, - child->last_siginfo); - case PTRACE_SETSIGINFO: - if (child->last_siginfo == NULL) - return -EINVAL; - return copy_siginfo_from_user32(child->last_siginfo, - (compat_siginfo_t - __force __user *) data); } - return ptrace_request(child, request, addr, data); + return compat_ptrace_request(child, request, addr, data); } #endif -long arch_ptrace(struct task_struct *child, long request, long addr, long data) -{ - switch (request) { - case PTRACE_SYSCALL: - /* continue and stop at next (return from) syscall */ - case PTRACE_CONT: - /* restart after signal. */ - if (!valid_signal(data)) - return -EIO; - if (request == PTRACE_SYSCALL) - set_tsk_thread_flag(child, TIF_SYSCALL_TRACE); - else - clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE); - child->exit_code = data; - /* make sure the single step bit is not set. */ - user_disable_single_step(child); - wake_up_process(child); - return 0; - - case PTRACE_KILL: - /* - * make the child exit. Best I can do is send it a sigkill. - * perhaps it should be put in the status that it wants to - * exit. - */ - if (child->exit_state == EXIT_ZOMBIE) /* already dead */ - return 0; - child->exit_code = SIGKILL; - /* make sure the single step bit is not set. */ - user_disable_single_step(child); - wake_up_process(child); - return 0; - - case PTRACE_SINGLESTEP: - /* set the trap flag. */ - if (!valid_signal(data)) - return -EIO; - clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE); - child->exit_code = data; - user_enable_single_step(child); - /* give it a chance to run. */ - wake_up_process(child); - return 0; - - /* Do requests that differ for 31/64 bit */ - default: -#ifdef CONFIG_COMPAT - if (test_thread_flag(TIF_31BIT)) - return do_ptrace_emu31(child, request, addr, data); -#endif - return do_ptrace_normal(child, request, addr, data); - } - /* Not reached. */ - return -EIO; -} - asmlinkage void syscall_trace(struct pt_regs *regs, int entryexit) { diff --git a/include/asm-s390/ptrace.h b/include/asm-s390/ptrace.h index 441d7c260857..d7d4e2eb3e6f 100644 --- a/include/asm-s390/ptrace.h +++ b/include/asm-s390/ptrace.h @@ -471,6 +471,8 @@ struct task_struct; extern void user_enable_single_step(struct task_struct *); extern void user_disable_single_step(struct task_struct *); +#define __ARCH_WANT_COMPAT_SYS_PTRACE + #define user_mode(regs) (((regs)->psw.mask & PSW_MASK_PSTATE) != 0) #define instruction_pointer(regs) ((regs)->psw.addr & PSW_ADDR_INSN) #define regs_return_value(regs)((regs)->gprs[2]) From 74c76c84576eb2d806f40f6cb2fc8302c01869d8 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Wed, 7 May 2008 09:22:58 +0200 Subject: [PATCH 7/8] [S390] tty3270: fix put_char fail/success conversion. The wrong function got coverted ;) CC drivers/s390/char/tty3270.o drivers/s390/char/tty3270.c:1747: warning: initialization from incompatible pointer type Acked-by: Alan Cox Cc: Peter Oberparleiter Signed-off-by: Heiko Carstens Signed-off-by: Martin Schwidefsky --- drivers/s390/char/tty3270.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) diff --git a/drivers/s390/char/tty3270.c b/drivers/s390/char/tty3270.c index c1f2adefad41..5043150019ac 100644 --- a/drivers/s390/char/tty3270.c +++ b/drivers/s390/char/tty3270.c @@ -965,8 +965,7 @@ tty3270_write_room(struct tty_struct *tty) * Insert character into the screen at the current position with the * current color and highlight. This function does NOT do cursor movement. */ -static int -tty3270_put_character(struct tty3270 *tp, char ch) +static void tty3270_put_character(struct tty3270 *tp, char ch) { struct tty3270_line *line; struct tty3270_cell *cell; @@ -986,7 +985,6 @@ tty3270_put_character(struct tty3270 *tp, char ch) cell->character = tp->view.ascebc[(unsigned int) ch]; cell->highlight = tp->highlight; cell->f_color = tp->f_color; - return 1; } /* @@ -1612,16 +1610,15 @@ tty3270_write(struct tty_struct * tty, /* * Put single characters to the ttys character buffer */ -static void -tty3270_put_char(struct tty_struct *tty, unsigned char ch) +static int tty3270_put_char(struct tty_struct *tty, unsigned char ch) { struct tty3270 *tp; tp = tty->driver_data; - if (!tp) - return; - if (tp->char_count < TTY3270_CHAR_BUF_SIZE) - tp->char_buf[tp->char_count++] = ch; + if (!tp || tp->char_count >= TTY3270_CHAR_BUF_SIZE) + return 0; + tp->char_buf[tp->char_count++] = ch; + return 1; } /* From 45e576b1c3d0020607b8666c0247164e92c7d719 Mon Sep 17 00:00:00 2001 From: Martin Schwidefsky Date: Wed, 7 May 2008 09:22:59 +0200 Subject: [PATCH 8/8] [S390] guest page hinting light Use the existing arch_alloc_page/arch_free_page callbacks to do the guest page state transitions between stable and unused. Acked-by: Rik van Riel Signed-off-by: Martin Schwidefsky --- arch/s390/Kconfig | 7 ++++ arch/s390/mm/Makefile | 1 + arch/s390/mm/init.c | 3 ++ arch/s390/mm/page-states.c | 79 ++++++++++++++++++++++++++++++++++++++ include/asm-s390/page.h | 11 ++++++ include/asm-s390/system.h | 6 +++ 6 files changed, 107 insertions(+) create mode 100644 arch/s390/mm/page-states.c diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig index 29a7940f284f..1d035082e78e 100644 --- a/arch/s390/Kconfig +++ b/arch/s390/Kconfig @@ -430,6 +430,13 @@ config CMM_IUCV Select this option to enable the special message interface to the cooperative memory management. +config PAGE_STATES + bool "Unused page notification" + help + This enables the notification of unused pages to the + hypervisor. The ESSA instruction is used to do the states + changes between a page that has content and the unused state. + config VIRT_TIMER bool "Virtual CPU timer support" help diff --git a/arch/s390/mm/Makefile b/arch/s390/mm/Makefile index fb988a48a754..2a7458134544 100644 --- a/arch/s390/mm/Makefile +++ b/arch/s390/mm/Makefile @@ -5,3 +5,4 @@ obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o obj-$(CONFIG_CMM) += cmm.o obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o +obj-$(CONFIG_PAGE_STATES) += page-states.o diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c index fa31de6ae97a..29f3a63806b9 100644 --- a/arch/s390/mm/init.c +++ b/arch/s390/mm/init.c @@ -126,6 +126,9 @@ void __init mem_init(void) /* clear the zero-page */ memset(empty_zero_page, 0, PAGE_SIZE); + /* Setup guest page hinting */ + cmma_init(); + /* this will put all low memory onto the freelists */ totalram_pages += free_all_bootmem(); diff --git a/arch/s390/mm/page-states.c b/arch/s390/mm/page-states.c new file mode 100644 index 000000000000..fc0ad73ffd90 --- /dev/null +++ b/arch/s390/mm/page-states.c @@ -0,0 +1,79 @@ +/* + * arch/s390/mm/page-states.c + * + * Copyright IBM Corp. 2008 + * + * Guest page hinting for unused pages. + * + * Author(s): Martin Schwidefsky + */ + +#include +#include +#include +#include +#include + +#define ESSA_SET_STABLE 1 +#define ESSA_SET_UNUSED 2 + +static int cmma_flag; + +static int __init cmma(char *str) +{ + char *parm; + parm = strstrip(str); + if (strcmp(parm, "yes") == 0 || strcmp(parm, "on") == 0) { + cmma_flag = 1; + return 1; + } + cmma_flag = 0; + if (strcmp(parm, "no") == 0 || strcmp(parm, "off") == 0) + return 1; + return 0; +} + +__setup("cmma=", cmma); + +void __init cmma_init(void) +{ + register unsigned long tmp asm("0") = 0; + register int rc asm("1") = -EOPNOTSUPP; + + if (!cmma_flag) + return; + asm volatile( + " .insn rrf,0xb9ab0000,%1,%1,0,0\n" + "0: la %0,0\n" + "1:\n" + EX_TABLE(0b,1b) + : "+&d" (rc), "+&d" (tmp)); + if (rc) + cmma_flag = 0; +} + +void arch_free_page(struct page *page, int order) +{ + int i, rc; + + if (!cmma_flag) + return; + for (i = 0; i < (1 << order); i++) + asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0" + : "=&d" (rc) + : "a" ((page_to_pfn(page) + i) << PAGE_SHIFT), + "i" (ESSA_SET_UNUSED)); +} + +void arch_alloc_page(struct page *page, int order) +{ + int i, rc; + + if (!cmma_flag) + return; + for (i = 0; i < (1 << order); i++) + asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0" + : "=&d" (rc) + : "a" ((page_to_pfn(page) + i) << PAGE_SHIFT), + "i" (ESSA_SET_STABLE)); +} diff --git a/include/asm-s390/page.h b/include/asm-s390/page.h index f0f4579eac13..12fd9c4f0f15 100644 --- a/include/asm-s390/page.h +++ b/include/asm-s390/page.h @@ -125,6 +125,17 @@ page_get_storage_key(unsigned long addr) return skey; } +#ifdef CONFIG_PAGE_STATES + +struct page; +void arch_free_page(struct page *page, int order); +void arch_alloc_page(struct page *page, int order); + +#define HAVE_ARCH_FREE_PAGE +#define HAVE_ARCH_ALLOC_PAGE + +#endif + #endif /* !__ASSEMBLY__ */ /* to align the pointer to the (next) page boundary */ diff --git a/include/asm-s390/system.h b/include/asm-s390/system.h index c819ae25a842..e0d4500d5f95 100644 --- a/include/asm-s390/system.h +++ b/include/asm-s390/system.h @@ -116,6 +116,12 @@ extern void pfault_fini(void); #define pfault_fini() do { } while (0) #endif /* CONFIG_PFAULT */ +#ifdef CONFIG_PAGE_STATES +extern void cmma_init(void); +#else +static inline void cmma_init(void) { } +#endif + #define finish_arch_switch(prev) do { \ set_fs(current->thread.mm_segment); \ account_vtime(prev); \