latest fixes for s390.
Subsystems (3)
cover latest fixes for s390.
Five new bug fixes for s390:
Christian Borntraeger (2):
[S390] vmcp: remove BKL
[S390] pgtables: Fix race in enable_sie vs. page table ops
Heiko Carstens (1):
[S390] Fix sysdev class file creation.
Jan Glauber (2):
[S390] qdio: prevent double qdio shutdown in case of I/O errors [S390] qdio: remove incorrect memset
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.
PATCH 1/5 qdio: prevent double qdio shutdown in case of I/O errors
From: Jan Glauber <[email protected]>
In case of I/O errors on a qdio subchannel qdio_shutdown may be called twice by the qdio driver and by zfcp. Remove the superfluous shutdown from qdio and let the upper layer driver handle the error condition.
---
drivers/s390/cio/qdio_main.c | 1 - 1 file changed, 1 deletion(-)
drivers/s390/cio/qdio_main.c +0 -1
@@ -1083,7 +1083,6 @@ void qdio_int_handler(struct ccw_devicecase -EIO :sprintf (dbf_text,"ierr%4x" , irq_ptr->schid.sch_no);QDIO_DBF_TEXT2 (1 , setup, dbf_text); -qdio_int_error (cdev);return ;case -ETIMEDOUT :sprintf (dbf_text,"qtoh%4x" , irq_ptr->schid.sch_no);
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.
PATCH 2/5 qdio: remove incorrect memset
In reply to Martin SchwidefskyFrom: Jan Glauber <[email protected]>
Remove the memset since zeroing the string is not needed and use snprintf instead of sprintf.
---
drivers/s390/cio/qdio_debug.c | 19 ++++++------------- 1 file changed, 6 insertions(+), 13 deletions(-)
drivers/s390/cio/qdio_debug.c +6 -13
@@ -20,6 +20,7 @@ static struct dentry *debugfs_root;#define MAX_DEBUGFS_QUEUES 32static struct dentry *debugfs_queues[MAX_DEBUGFS_QUEUES ] = {NULL };static DEFINE_MUTEX (debugfs_mutex); +#define QDIO_DEBUGFS_NAME_LEN 40void qdio_allocate_do_dbf (struct qdio_initialize *init_data) {
@@ -152,17 +153,6 @@ static int qstat_seq_open(struct inode * filp->f_path.dentry->d_inode->i_private); } -static void get_queue_name (struct qdio_q *q,struct ccw_device *cdev,char *name) -{ -memset (name,0 ,sizeof (name)); -sprintf (name,"%s" ,dev_name (&cdev->dev)); -if (q->is_input_q) -sprintf (name +strlen (name),"_input" ); -else -sprintf (name +strlen (name),"_output" ); -sprintf (name +strlen (name),"_%d" , q->nr); -} -static void remove_debugfs_entry (struct qdio_q *q) {int i;
@@ -189,14 +179,17 @@ static struct file_operations debugfs_fostatic void setup_debugfs_entry (struct qdio_q *q,struct ccw_device *cdev) {int i =0 ; -char name[40 ]; +char name[QDIO_DEBUGFS_NAME_LEN ];while (debugfs_queues[i] !=NULL ) { i++;if (i >=MAX_DEBUGFS_QUEUES )return ; } -get_queue_name (q, cdev, name); +snprintf (name,QDIO_DEBUGFS_NAME_LEN ,"%s_%s_%d" , +dev_name (&cdev->dev), + q->is_input_q ?"input" :"output" , + q->nr); debugfs_queues[i] =debugfs_create_file (name,S_IFREG |S_IRUGO |S_IWUSR , debugfs_root, q, &debugfs_fops); }
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.
PATCH 3/5 vmcp: remove BKL
In reply to Martin SchwidefskyFrom: Christian Borntraeger <[email protected]>
The vmcp driver uses the session->mutex for concurrent access of the data structures. Therefore, the BKL in vmcp_open does not protect against any other function in the driver.
The BLK in vmcp_open would protect concurrent access to the module init but all necessary steps ave finished before misc_register is called. We can safely remove the lock_kernel from vcmp.
---
drivers/s390/char/vmcp.c | 3 --- 1 file changed, 3 deletions(-)
drivers/s390/char/vmcp.c +0 -3
@@ -16,7 +16,6 @@#include <linux/kernel.h> #include <linux/miscdevice.h> #include <linux/module.h> -#include <linux/smp_lock.h> #include <asm/cpcmd.h> #include <asm/debug.h> #include <asm/uaccess.h>
@@ -41,13 +40,11 @@ static int vmcp_open(struct inode *inodeif (!session)return -ENOMEM ; -lock_kernel (); session->bufsize =PAGE_SIZE ; session->response =NULL ; session->resp_size =0 ;mutex_init (&session->mutex); file->private_data = session; -unlock_kernel ();return nonseekable_open (inode, file); }
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.
PATCH 4/5 pgtables: Fix race in enable_sie vs. page table ops
In reply to Martin SchwidefskyFrom: Christian Borntraeger <[email protected]>
The current enable_sie code sets the mm->context.pgstes bit to tell dup_mm that the new mm should have extended page tables. This bit is also used by the s390 specific page table primitives to decide about the page table layout - which means context.pgstes has two meanings. This can cause any kind of bugs. For example - e.g. shrink_zone can call ptep_clear_flush_young while enable_sie is running. ptep_clear_flush_young will test for context.pgstes. Since enable_sie changed that value of the old struct mm without changing the page table layout ptep_clear_flush_young will do the wrong thing.
The solution is to split pgstes into two bits
- one for the allocation
- one for the current state
---
arch/s390/include/asm/mmu.h | 3 ++- arch/s390/include/asm/mmu_context.h | 19 ++++++++++++++++--- arch/s390/include/asm/pgtable.h | 8 ++++---- arch/s390/mm/pgtable.c | 16 ++++++++-------- 4 files changed, 30 insertions(+), 16 deletions(-)
arch/s390/include/asm/mmu_context.h +16 -3
@@ -20,12 +20,25 @@ static inline int init_new_context(struc#ifdef CONFIG_64BIT mm->context.asce_bits |= _ASCE_TYPE_REGION3;#endif -if (current->mm->context.pgstes) { +if (current->mm->context.alloc_pgste) { +/* +* alloc_pgste indicates, that any NEW context will be created +* with extended page tables. The old context is unchanged. The +* page table allocation and the page table operations will +* look at has_pgste to distinguish normal and extended page +* tables. The only way to create extended page tables is to +* set alloc_pgste and then create a new context (e.g. dup_mm). +* The page table allocation is called after init_new_context +* and if has_pgste is set, it will create extended page +* tables. +*/ mm->context.noexec =0 ; - mm->context.pgstes =1 ; + mm->context.has_pgste =1 ; + mm->context.alloc_pgste =1 ; }else { mm->context.noexec = s390_noexec; - mm->context.pgstes =0 ; + mm->context.has_pgste =0 ; + mm->context.alloc_pgste =0 ; } mm->context.asce_limit =STACK_TOP_MAX ;crst_table_init ((unsigned long *) mm->pgd,pgd_entry_type (mm));
arch/s390/include/asm/mmu.h +2 -1
@@ -7,7 +7,8 @@ typedef struct {unsigned long asce_bits;unsigned long asce_limit;int noexec; - int pgstes; +int has_pgste;/* The mmu context has extended page tables */ + int alloc_pgste; /* cloned contexts will have extended page tables */ } mm_context_t;#endif
arch/s390/include/asm/pgtable.h +4 -4
@@ -679,7 +679,7 @@ static inline void pmd_clear(pmd_t *pmd)static inline void pte_clear (struct mm_struct *mm,unsigned long addr,pte_t *ptep) { -if (mm->context.pgstes) +if (mm->context.has_pgste)ptep_rcp_copy (ptep);pte_val (*ptep) = _PAGE_TYPE_EMPTY;if (mm->context.noexec)
@@ -763,7 +763,7 @@ static inline int kvm_s390_test_and_cleastruct page *page;unsigned int skey; -if (!mm->context.pgstes) +if (!mm->context.has_pgste)return -EINVAL ;rcp_lock (ptep); pgste = (unsigned long *) (ptep +PTRS_PER_PTE );
@@ -794,7 +794,7 @@ static inline int ptep_test_and_clear_yoint young;unsigned long *pgste; -if (!vma->vm_mm->context.pgstes) +if (!vma->vm_mm->context.has_pgste)return 0 ; physpage =pte_val (*ptep) &PAGE_MASK ; pgste = (unsigned long *) (ptep +PTRS_PER_PTE );
@@ -844,7 +844,7 @@ static inline void __ptep_ipte(unsignedstatic inline void ptep_invalidate (struct mm_struct *mm,unsigned long address,pte_t *ptep) { -if (mm->context.pgstes) { +if (mm->context.has_pgste) {rcp_lock (ptep);__ptep_ipte (address, ptep);ptep_rcp_copy (ptep);
arch/s390/mm/pgtable.c +8 -8
@@ -169,7 +169,7 @@ unsigned long *page_table_alloc(struct munsigned long *table;unsigned long bits; - bits = (mm->context.noexec || mm->context.pgstes) ?3UL :1UL ; + bits = (mm->context.noexec || mm->context.has_pgste) ?3UL :1UL ;spin_lock (&mm->page_table_lock); page =NULL ;if (!list_empty (&mm->context.pgtable_list)) {
@@ -186,7 +186,7 @@ unsigned long *page_table_alloc(struct mpgtable_page_ctor (page); page->flags &= ~FRAG_MASK ; table = (unsigned long *)page_to_phys (page); -if (mm->context.pgstes) +if (mm->context.has_pgste)clear_table_pgstes (table);else clear_table (table, _PAGE_TYPE_EMPTY,PAGE_SIZE );
@@ -210,7 +210,7 @@ void page_table_free(struct mm_struct *mstruct page *page;unsigned long bits; - bits = (mm->context.noexec || mm->context.pgstes) ?3UL :1UL ; + bits = (mm->context.noexec || mm->context.has_pgste) ?3UL :1UL ; bits <<= (__pa (table) & (PAGE_SIZE -1 )) /256 /sizeof (unsigned long ); page =pfn_to_page (__pa (table) >>PAGE_SHIFT );spin_lock (&mm->page_table_lock);
@@ -257,7 +257,7 @@ int s390_enable_sie(void)struct mm_struct *mm, *old_mm;/* Do we have pgstes? if yes, we are done */ -if (tsk->mm->context.pgstes) +if (tsk->mm->context.has_pgste)return 0 ;/* lets check if we are allowed to replace the mm */
@@ -269,14 +269,14 @@ int s390_enable_sie(void) }task_unlock (tsk); -/* we copy the mm with pgstes enabled */ - tsk->mm->context.pgstes =1 ; +/* we copy the mm and let dup_mm create the page tables with_pgstes */ + tsk->mm->context.alloc_pgste =1 ; mm =dup_mm (tsk); - tsk->mm->context.pgstes =0 ; + tsk->mm->context.alloc_pgste =0 ;if (!mm)return -ENOMEM ; -/* Now lets check again if somebody attached ptrace etc */ +/* Now lets check again if something happened */ task_lock (tsk);if (!tsk->mm ||atomic_read (&tsk->mm->mm_users) >1 || tsk->mm != tsk->active_mm || tsk->mm->ioctx_list) {
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.
PATCH 5/5 Fix sysdev class file creation.
In reply to Martin SchwidefskyFrom: Heiko Carstens <[email protected]>
Use sysdev_class_create_file() to create create sysdev class attributes instead of sysfs_create_file(). Using sysfs_create_file() wasn't a very good idea since the show and store functions have a different amount of parameters for sysfs files and sysdev class files.
In particular the pointer to the buffer is the last argument and therefore accesses to random memory regions happened.
Still worked surprisingly well until we got a kernel panic.
---
arch/s390/kernel/smp.c | 24 +++++++++--------------- 1 file changed, 9 insertions(+), 15 deletions(-)
arch/s390/kernel/smp.c +9 -15
@@ -1119,9 +1119,7 @@ out:return rc; } -static ssize_t __ref rescan_store (struct sys_device *dev, -struct sysdev_attribute *attr, -const char *buf, +static ssize_t __ref rescan_store (struct sysdev_class *class,const char *buf,size_t count) {int rc;
@@ -1129,12 +1127,10 @@ static ssize_t __ref rescan_store(struct rc =smp_rescan_cpus ();return rc ? rc : count; } -static SYSDEV_ATTR(rescan, 0200, NULL, rescan_store); +static SYSDEV_CLASS_ATTR(rescan, 0200, NULL, rescan_store);#endif /* CONFIG_HOTPLUG_CPU */ -static ssize_t dispatching_show (struct sys_device *dev, -struct sysdev_attribute *attr, -char *buf) +static ssize_t dispatching_show (struct sysdev_class *class,char *buf) {ssize_t count;
@@ -1144,9 +1140,8 @@ static ssize_t dispatching_show(struct sreturn count; } -static ssize_t dispatching_store (struct sys_device *dev, -struct sysdev_attribute *attr, -const char *buf,size_t count) +static ssize_t dispatching_store (struct sysdev_class *dev,const char *buf, +size_t count) {int val, rc;char delim;
@@ -1168,7 +1163,8 @@ out:put_online_cpus ();return rc ? rc : count; } -static SYSDEV_ATTR(dispatching, 0644, dispatching_show, dispatching_store); +static SYSDEV_CLASS_ATTR(dispatching, 0644, dispatching_show, + dispatching_store);static int __init topology_init (void ) {
@@ -1178,13 +1174,11 @@ static int __init topology_init(void)register_cpu_notifier (&smp_cpu_nb);#ifdef CONFIG_HOTPLUG_CPU - rc =sysfs_create_file (&cpu_sysdev_class.kset.kobj, - &attr_rescan.attr); + rc =sysdev_class_create_file (&cpu_sysdev_class, &attr_rescan);if (rc)return rc;#endif - rc =sysfs_create_file (&cpu_sysdev_class.kset.kobj, - &attr_dispatching.attr); + rc =sysdev_class_create_file (&cpu_sysdev_class, &attr_dispatching);if (rc)return rc;for_each_present_cpu (cpu) {
Signature · 4 lines
blue skies, Martin. "Reality continues to ruin my life." - Calvin.