remove BKL from drivers' release functions
Subsystems (14)
PATCH 1/1 remove BKL from drivers' release functions
This version fixes a non-release of a spinlock before a return on error in rd.c.
The following is a patch which removes the BKL from quite a few drivers' release functions. The release functions are already serialized in the VFS code by an atomic_t which guarantees that each function will be called only once, after all file descriptors have been closed. In addition, in these drivers, the BKL was _only_ held in the release function and nowhere else in the driver where it might be needed.
Many of these patches simply remove the BKL from the file. This causes no harm because the BKL was not really protecting anything, anyway. Other patches try to actually fix the locking. Some do this by making use of atomic operations with the atomic_* functions, or the (test|set)_bit functions. Most of these patches replace uses of normal integers which were used to keep open counts in the drivers. In other some cases, a spinlock was added when the atomic operations could not guarantee proper serialization by themselves. And, in very few cases, the existing locking was extended to protect more things. These cases are very uncommon because locking is very uncommon in most of these drivers.
Special care has been taken not to introduce more locking issues into the drivers (do no harm).
The patch applies cleanly against 2.5.1-pre2.
arch/i386/kernel/apm.c +9 -2
@@ -387,6 +387,7 @@static DECLARE_WAIT_QUEUE_HEAD (apm_waitqueue);static DECLARE_WAIT_QUEUE_HEAD (apm_suspend_waitqueue);static struct apm_user * user_list; +static spinlock_t user_list_lock;static char driver_version[] ="1.15" ;/* no spaces */
@@ -1053,6 +1054,7 @@ {struct apm_user * as; +spin_lock ( &user_list_lock );if (user_list ==NULL )return ;for (as = user_list; as !=NULL ; as = as->next) {
@@ -1083,6 +1085,7 @@break ; } } +spin_unlock ( &user_list_lock );wake_up_interruptible (&apm_waitqueue); }
@@ -1179,10 +1182,12 @@send_event (APM_NORMAL_RESUME );sti ();queue_event (APM_NORMAL_RESUME ,NULL ); +spin_lock ( &user_list_lock );for (as = user_list; as !=NULL ; as = as->next) { as->suspend_wait =0 ; as->suspend_result = ((err ==APM_SUCCESS ) ?0 : -EIO ); } +spin_unlock ( &user_list_lock ); ignore_normal_resume =1 ;wake_up_interruptible (&apm_suspend_waitqueue);return err;
@@ -1519,7 +1524,6 @@if (check_apm_user (as,"release" ))return 0 ; filp->private_data =NULL ; -lock_kernel ();if (as->standbys_pending >0 ) { standbys_pending -= as->standbys_pending;if (standbys_pending <=0 )
@@ -1530,6 +1534,7 @@if (suspends_pending <=0 ) (void )suspend (); } +spin_lock ( &user_list_lock );if (user_list == as) user_list = as->next;else {
@@ -1544,7 +1549,7 @@else as1->next = as->next; } -unlock_kernel (); +spin_unlock ( &user_list_lock );kfree (as);return 0 ; }
@@ -1573,8 +1578,10 @@ as->suser =capable (CAP_SYS_ADMIN ); as->writer = (filp->f_mode &FMODE_WRITE ) ==FMODE_WRITE ; as->reader = (filp->f_mode &FMODE_READ ) ==FMODE_READ ; +spin_lock ( &user_list_lock ); as->next = user_list; user_list = as; +spin_unlock ( &user_list_lock ); filp->private_data = as;return 0 ; }
arch/i386/kernel/mtrr.c +0 -2
@@ -1788,7 +1788,6 @@unsigned int *fcount = file->private_data;if (fcount ==NULL )return 0 ; -lock_kernel (); max =get_num_var_ranges ();for (i =0 ; i < max; ++i) {
@@ -1798,7 +1797,6 @@ --fcount[i]; } } -unlock_kernel ();kfree (fcount); file->private_data =NULL ;return 0 ;
arch/m68k/atari/joystick.c +0 -2
@@ -61,13 +61,11 @@ {int minor =DEVICE_NR (inode->i_rdev); -lock_kernel (); joystick[minor].active =0 ; joystick[minor].ready =0 ;if ((joystick[0 ].active ==0 ) && (joystick[1 ].active ==0 ))ikbd_joystick_disable (); -unlock_kernel ();return 0 ; }
arch/m68k/mvme16x/rtc.c +6 -6
@@ -36,7 +36,7 @@static unsigned char days_in_mo[] = {0 ,31 ,28 ,31 ,30 ,31 ,30 ,31 ,31 ,30 ,31 ,30 ,31 }; -static char rtc_status =0 ; +static atomic_t rtc_ready =ATOMIC_INIT (1 );static int rtc_ioctl (struct inode *inode,struct file *file,unsigned int cmd,unsigned long arg)
@@ -130,18 +130,18 @@static int rtc_open (struct inode *inode,struct file *file) { -if (rtc_status) +if ( !atomic_dec_and_test (&rtc_ready) ) + { +atomic_inc ( &rtc_ready );return -EBUSY ; + } - rtc_status =1 ;return 0 ; }static int rtc_release (struct inode *inode,struct file *file) { -lock_kernel (); - rtc_status =0 ; -unlock_kernel (); +atomic_inc ( &rtc_ready );return 0 ; }
arch/mips64/sgi-ip27/ip27-rtc.c +6 -19
@@ -53,14 +53,7 @@static void get_rtc_time (struct rtc_time *rtc_tm); -/* -* Bits in rtc_status. (6 bits of room for future expansion) -*/ - -#define RTC_IS_OPEN 0x01/* means /dev/rtc is in use */ -#define RTC_TIMER_ON 0x02/* missed irq timer active */ - -static unsigned char rtc_status;/* bitmapped status byte. */ +static atomic_t rtc_ready =ATOMIC_INIT (1 );static unsigned long rtc_freq;/* Current periodic IRQ rate */ static struct m48t35_rtc *rtc;
@@ -166,23 +159,17 @@static int rtc_open (struct inode *inode,struct file *file) { -if (rtc_status &RTC_IS_OPEN ) +if (atomic_dec_and_test ( &rtc_ready ) ) + { +atomic_inc ( &rtc_ready );return -EBUSY ; - - rtc_status |=RTC_IS_OPEN ; + }return 0 ; }static int rtc_release (struct inode *inode,struct file *file) { -/* -* Turn off all interrupts once the device is no longer -* in use, and clear the data. -*/ - -lock_kernel (); - rtc_status &= ~RTC_IS_OPEN ; -unlock_kernel (); +atomic_inc ( &rtc_ready );return 0 ; }
arch/sparc64/solaris/socksys.c +0 -2
@@ -118,7 +118,6 @@struct T_primsg *it;/* XXX: check this */ -lock_kernel (); sock = (struct sol_socket_struct *)filp->private_data;SOLDD (("sock release %016lx(%016lx)\n" , sock, filp)); it = sock->pfirst;
@@ -132,7 +131,6 @@ filp->private_data =NULL ;SOLDD (("socksys_release %016lx\n" , sock));mykfree ((char *)sock); -unlock_kernel ();return 0 ; }
drivers/block/acsi_slm.c +14 -14
@@ -121,8 +121,8 @@static struct slm {unsigned target;/* target number */ unsigned lun;/* LUN in target controller */ -unsigned wbusy :1 ;/* output part busy */ -unsigned rbusy :1 ;/* status part busy */ +atomic_t wr_ok;/* set to 0 if output part busy */ +atomic_t rd_ok;/* set to 0 if status part busy */ } slm_info[MAX_SLM ];int N_SLM_Printers =0 ;
@@ -778,15 +778,17 @@if (file->f_mode &2 ) {/* open for writing is exclusive */ -if (sip->wbusy) +if ( !atomic_dec_and_test (&sip->wr_ok) ) { +atomic_inc (&sip->wr_ok);return ( -EBUSY ); - sip->wbusy =1 ; + } }if (file->f_mode &1 ) { -/* open for writing is exclusive */ -if (sip->rbusy) -return ( -EBUSY ); - sip->rbusy =1 ; +/* open for reading is exclusive */ +if ( !atomic_dec_and_test (&sip->rd_ok) ) { +atomic_inc (&sip->rd_ok); +return ( -EBUSY ); + } }return (0 );
@@ -801,12 +803,10 @@ device =MINOR (inode->i_rdev); sip = &slm_info[device]; -lock_kernel ();if (file->f_mode &2 ) - sip->wbusy =0 ; +atomic_inc ( &sip->wr_ok );if (file->f_mode &1 ) - sip->rbusy =0 ; -unlock_kernel (); +atomic_inc ( &sip->rd_ok );return (0 ); }
@@ -983,8 +983,8 @@ slm_info[N_SLM_Printers].target = target; slm_info[N_SLM_Printers].lun = lun; - slm_info[N_SLM_Printers].wbusy =0 ; - slm_info[N_SLM_Printers].rbusy =0 ; +atomic_set (&slm_info[N_SLM_Printers].wr_ok,1 ); +atomic_set (&slm_info[N_SLM_Printers].rd_ok,1 );printk (KERN_INFO " Printer: %s\n" , SLMBuffer ); printk( KERN_INFO "Detected slm%d at id %d lun %d\n",
drivers/block/paride/pg.c +5 -10
@@ -276,7 +276,7 @@ pg_drive_count =0 ;for (unit=0 ;unit<PG_UNITS ;unit++) {PG .pi = &PG .pia; -PG .access =0 ; +set_bit (0 , &PG .access );PG .busy =0 ;PG .present =0 ;PG .bufptr =NULL ;
@@ -573,10 +573,7 @@if ((unit >=PG_UNITS ) || (!PG .present))return -ENODEV ; -PG .access++; - -if (PG .access >1 ) { -PG .access--; +if (test_and_set_bit (0 , &PG .access) ) {return -EBUSY ; }
@@ -590,7 +587,7 @@PG .bufptr =kmalloc (PG_MAX_DATA ,GFP_KERNEL );if (PG .bufptr ==NULL ) { -PG .access--; +clear_bit (0 , &PG .access ) ;printk ("%s: buffer allocation failed\n" ,PG .name);return -ENOMEM ; }
@@ -602,15 +599,13 @@ {int unit =DEVICE_NR (inode->i_rdev); -if ((unit >=PG_UNITS ) || (PG .access <=0 )) +if ( unit >=PG_UNITS || !test_bit (0 ,&PG .access) )return -EINVAL ; -lock_kernel (); -PG .access--; +clear_bit (0 , &PG .access);kfree (PG .bufptr);PG .bufptr =NULL ; -unlock_kernel ();return 0 ;
drivers/block/paride/pt.c +10 -14
@@ -244,7 +244,7 @@int flags;/* various state flags */ int last_sense;/* result of last request sense */ int drive;/* drive */ -int access;/* count of active opens ... */ +atomic_t available;/* 1 if access is available 0 otherwise */ int bs;/* block size */ int capacity;/* Size of tape in KB */ int present;/* device present ? */
@@ -279,7 +279,7 @@ pt_drive_count =0 ;for (unit=0 ;unit<PT_UNITS ;unit++) {PT .pi = &PT .pia; -PT .access =0 ; +atomic_set ( &PT .available,1 );PT .flags =0 ;PT .last_sense =0 ;PT .present =0 ;
@@ -696,22 +696,20 @@if ((unit >=PT_UNITS ) || (!PT .present))return -ENODEV ; -PT .access++; - -if (PT .access >1 ) { -PT .access--; +if ( !atomic_dec_and_test (&PT .available) ) { +atomic_inc ( &PT .available );return -EBUSY ; }pt_identify (unit);if (!PT .flags &PT_MEDIA ) { -PT .access--; +atomic_inc ( &PT .available );return -ENODEV ; }if ((!PT .flags &PT_WRITE_OK ) && (file ->f_mode &2 )) { -PT .access--; +atomic_inc ( &PT .available );return -EROFS ; }
@@ -720,7 +718,7 @@PT .bufptr =kmalloc (PT_BUFSIZE ,GFP_KERNEL );if (PT .bufptr ==NULL ) { -PT .access--; +atomic_inc ( &PT .available );printk ("%s: buffer allocation failed\n" ,PT .name);return -ENOMEM ; }
@@ -775,20 +773,18 @@ {int unit =DEVICE_NR (inode->i_rdev); -if ((unit >=PT_UNITS ) || (PT .access <=0 )) +if ((unit >=PT_UNITS ) || (atomic_read (&PT .available) >1 ))return -EINVAL ; -lock_kernel ();if (PT .flags &PT_WRITING )pt_write_fm (unit);if (PT .flags &PT_REWIND )pt_rewind (unit); -PT .access--; -kfree (PT .bufptr);PT .bufptr =NULL ; -unlock_kernel (); +atomic_inc ( &PT .available ); +return 0 ; }
drivers/block/rd.c +9 -3
@@ -89,6 +89,7 @@#ifdef CONFIG_BLK_DEV_INITRD static int initrd_users; +static spinlock_t initrd_users_lock =SPIN_LOCK_UNLOCKED ;#endif #endif
@@ -408,12 +409,15 @@ {extern void free_initrd_mem (unsigned long ,unsigned long ); -lock_kernel (); +spin_lock ( &initrd_users_lock );if (!--initrd_users) { +spin_unlock ( &initrd_users_lock );free_initrd_mem (initrd_start, initrd_end); initrd_start =0 ; + }else { +spin_unlock ( &initrd_users_lock ); } -unlock_kernel (); +blkdev_put (inode->i_bdev,BDEV_FILE );return 0 ; }
@@ -433,8 +437,10 @@#ifdef CONFIG_BLK_DEV_INITRD if (unit ==INITRD_MINOR ) { -if (!initrd_start)return -ENODEV ; +spin_lock ( &initrd_users_lock ); initrd_users++; +if (!initrd_start)return -ENODEV ; +spin_unlock ( &initrd_users_lock ); filp->f_op = &initrd_fops;return 0 ; }
drivers/char/acquirewdt.c +0 -2
@@ -141,7 +141,6 @@static int acq_close (struct inode *inode,struct file *file) { -lock_kernel ();if (MINOR (inode->i_rdev)==WATCHDOG_MINOR ) {spin_lock (&acq_lock);
@@ -151,7 +150,6 @@ acq_is_open=0 ;spin_unlock (&acq_lock); } -unlock_kernel ();return 0 ; }
drivers/char/advantechwdt.c +0 -2
@@ -151,7 +151,6 @@static int advwdt_close (struct inode *inode,struct file *file) { -lock_kernel ();if (MINOR (inode->i_rdev) ==WATCHDOG_MINOR ) {spin_lock (&advwdt_lock);#ifndef CONFIG_WATCHDOG_NOWAYOUT
@@ -160,7 +159,6 @@ advwdt_is_open =0 ;spin_unlock (&advwdt_lock); } -unlock_kernel ();return 0 ; }
drivers/char/agp/agpgart_fe.c +0 -14
@@ -606,17 +606,14 @@agp_file_private *priv = (agp_file_private *) file->private_data;agp_kern_info kerninfo; -lock_kernel ();AGP_LOCK ();if (agp_fe.backend_acquired != TRUE) {AGP_UNLOCK (); -unlock_kernel ();return -EPERM ; }if (!(test_bit (AGP_FF_IS_VALID , &priv->access_flags))) {AGP_UNLOCK (); -unlock_kernel ();return -EPERM ; }agp_copy_info (&kerninfo);
@@ -628,51 +625,42 @@if (test_bit (AGP_FF_IS_CLIENT , &priv->access_flags)) {if ((size + offset) > current_size) {AGP_UNLOCK (); -unlock_kernel ();return -EINVAL ; } client =agp_find_client_by_pid (current->pid);if (client ==NULL ) {AGP_UNLOCK (); -unlock_kernel ();return -EPERM ; }if (!agp_find_seg_in_client (client, offset, size, vma->vm_page_prot)) {AGP_UNLOCK (); -unlock_kernel ();return -EINVAL ; }if (remap_page_range (vma->vm_start, (kerninfo.aper_base + offset), size, vma->vm_page_prot)) {AGP_UNLOCK (); -unlock_kernel ();return -EAGAIN ; }AGP_UNLOCK (); -unlock_kernel ();return 0 ; }if (test_bit (AGP_FF_IS_CONTROLLER , &priv->access_flags)) {if (size != current_size) {AGP_UNLOCK (); -unlock_kernel ();return -EINVAL ; }if (remap_page_range (vma->vm_start, kerninfo.aper_base, size, vma->vm_page_prot)) {AGP_UNLOCK (); -unlock_kernel ();return -EAGAIN ; }AGP_UNLOCK (); -unlock_kernel ();return 0 ; }AGP_UNLOCK (); -unlock_kernel ();return -EPERM ; }
@@ -680,7 +668,6 @@ {agp_file_private *priv = (agp_file_private *) file->private_data; -lock_kernel ();AGP_LOCK ();if (test_bit (AGP_FF_IS_CONTROLLER , &priv->access_flags)) {
@@ -702,7 +689,6 @@agp_remove_file_private (priv);kfree (priv);AGP_UNLOCK (); -unlock_kernel ();return 0 ; }
drivers/char/busmouse.c +3 -1
@@ -171,6 +171,7 @@lock_kernel ();busmouse_fasync (-1 , file,0 ); +down (&mouse_sem);/* to protect mse->active */ if (--mse->active ==0 ) {if (mse->ops->release) ret = mse->ops->release (inode, file);
@@ -179,7 +180,8 @@ mse->ready =0 ; }unlock_kernel (); - +up ( &mouse_sem); +return ret; }
drivers/char/dtlk.c +1 -3
@@ -328,10 +328,8 @@break ; }TRACE_RET ; - -lock_kernel (); +del_timer (&dtlk_timer); -unlock_kernel ();return 0 ; }
drivers/char/ftape/zftape/zftape-init.c +12 -17
@@ -68,7 +68,8 @@/* Local vars. */ -static int busy_flag =0 ; +static int busy_flag; +static sigset_t orig_sigmask;/* the interface to the kernel vfs layer
@@ -113,14 +114,13 @@TRACE_FUN (ft_t_flow);TRACE (ft_t_flow,"called for minor %d" ,MINOR (ino->i_rdev)); -if (busy_flag) { + if ( test_and_set_bit(0,&busy_flag) )) {TRACE_ABORT (-EBUSY , ft_t_warn,"failed: already busy" ); } - busy_flag =1 ;if ((MINOR (ino->i_rdev) & ~(ZFT_MINOR_OP_MASK |FTAPE_NO_REWIND )) >FTAPE_SEL_D ) { - busy_flag =0 ; +clear_bit (0 ,&busy_flag);TRACE_ABORT (-ENXIO , ft_t_err,"failed: illegal unit nr" ); } orig_sigmask = current->blocked;
@@ -128,7 +128,7 @@ result =_zft_open (MINOR (ino->i_rdev), filep->f_flags &O_ACCMODE );if (result <0 ) { current->blocked = orig_sigmask;/* restore mask */ - busy_flag =0 ; +clear_bit (0 ,&busy_flag);TRACE_ABORT (result, ft_t_err,"_ftape_open failed" ); }else {/* Mask signals that will disturb proper operation of the
@@ -147,10 +147,8 @@int result;TRACE_FUN (ft_t_flow); -lock_kernel (); -if (!busy_flag ||MINOR (ino->i_rdev) != zft_unit) { +if ( !test_bit (0 ,&busy_flag) ||MINOR (ino->i_rdev) != zft_unit) {TRACE (ft_t_err,"failed: not busy or wrong unit" ); -unlock_kernel (); TRACE_EXIT 0; }sigfillset (¤t->blocked);
@@ -159,8 +157,7 @@TRACE (ft_t_err,"_zft_close failed" ); } current->blocked = orig_sigmask;/* restore before open state */ - busy_flag =0 ; -unlock_kernel (); +clear_bit (0 ,&busy_flag); TRACE_EXIT 0; }
@@ -173,7 +170,7 @@sigset_t old_sigmask;TRACE_FUN (ft_t_flow); -if (!busy_flag ||MINOR (ino->i_rdev) != zft_unit || ft_failure) { +if ( !test_bit (0 ,&busy_flag) ||MINOR (ino->i_rdev) != zft_unit || ft_failure) {TRACE_ABORT (-EIO , ft_t_err,"failed: not busy, failure or wrong unit" ); }
@@ -193,7 +190,7 @@sigset_t old_sigmask;TRACE_FUN (ft_t_flow); -if (!busy_flag || +if ( !test_bit (0 ,&busy_flag) ||MINOR (filep->f_dentry->d_inode->i_rdev) != zft_unit || ft_failure) {
@@ -202,14 +199,12 @@ } old_sigmask = current->blocked;/* save mask */ sigfillset (¤t->blocked); -lock_kernel ();if ((result =ftape_mmap (vma)) >=0 ) {#ifndef MSYNC_BUG_WAS_FIXED static struct vm_operations_struct dummy = {NULL , }; vma->vm_ops = &dummy;#endif } -unlock_kernel (); current->blocked = old_sigmask;/* restore mask */ TRACE_EXIT result; }
@@ -225,7 +220,7 @@TRACE_FUN (ft_t_flow);TRACE (ft_t_data_flow,"called with count: %ld" , (unsigned long )req_len); -if (!busy_flag ||MINOR (ino->i_rdev) != zft_unit || ft_failure) { +if (!test_bit (0 ,&busy_flag) ||MINOR (ino->i_rdev) != zft_unit || ft_failure) {TRACE_ABORT (-EIO , ft_t_err,"failed: not busy, failure or wrong unit" ); }
@@ -248,7 +243,7 @@TRACE_FUN (ft_t_flow);TRACE (ft_t_flow,"called with count: %ld" , (unsigned long )req_len); -if (!busy_flag ||MINOR (ino->i_rdev) != zft_unit || ft_failure) { +if (!test_bit (0 ,&busy_flag) ||MINOR (ino->i_rdev) != zft_unit || ft_failure) {TRACE_ABORT (-EIO , ft_t_err,"failed: not busy, failure or wrong unit" ); }
@@ -403,7 +398,7 @@*/ static int can_unload (void ) { -return (GET_USE_COUNT (THIS_MODULE )||zft_dirty ()||busy_flag)?-EBUSY :0 ; +return (GET_USE_COUNT (THIS_MODULE )||zft_dirty ()||test_bit (0 ,&busy_flag))?-EBUSY :0 ; }/* Called by modules package when installing the driver */
drivers/char/lp.c +0 -2
@@ -494,11 +494,9 @@parport_negotiate (lp_table[minor].dev->port,IEEE1284_MODE_COMPAT ); lp_table[minor].current_mode =IEEE1284_MODE_COMPAT ;lp_release_parport (&lp_table[minor]); -lock_kernel ();kfree (lp_table[minor].lp_buffer); lp_table[minor].lp_buffer =NULL ;LP_F (minor) &= ~LP_BUSY ; -unlock_kernel ();return 0 ; }
drivers/char/mixcomwd.c +0 -3
@@ -101,11 +101,9 @@static int mixcomwd_release (struct inode *inode,struct file *file) { -lock_kernel ();#ifndef CONFIG_WATCHDOG_NOWAYOUT if (mixcomwd_timer_alive) {printk (KERN_ERR "mixcomwd: release called while internal timer alive" ); -unlock_kernel ();return -EBUSY ; }init_timer (&mixcomwd_timer);
@@ -117,7 +115,6 @@#endif clear_bit (0 ,&mixcomwd_opened); -unlock_kernel ();return 0 ; }
drivers/char/nvram.c +11 -4
@@ -108,7 +108,10 @@#include <asm/system.h> static int nvram_open_cnt;/* #times opened */ -static int nvram_open_mode;/* special open modes */ +static int nvram_open_mode;/* special open modes */ +static spinlock_t nvram_open_lock =SPIN_LOCK_UNLOCKED ; +/* guards nvram_open_cnt and +nvram_open_mode */ #define NVRAM_WRITE 1/* opened for writing (exclusive) */ #define NVRAM_EXCL 2/* opened with O_EXCL */
@@ -326,29 +329,33 @@static int nvram_open (struct inode *inode,struct file *file ) { +spin_lock ( &nvram_open_lock );if ((nvram_open_cnt && (file->f_flags &O_EXCL )) || (nvram_open_mode &NVRAM_EXCL ) || ((file->f_mode &2 ) && (nvram_open_mode &NVRAM_WRITE ))) + { +spin_unlock ( &nvram_open_lock );return ( -EBUSY ); + }if (file->f_flags &O_EXCL ) nvram_open_mode |=NVRAM_EXCL ;if (file->f_mode &2 ) nvram_open_mode |=NVRAM_WRITE ; nvram_open_cnt++; +spin_unlock ( &nvram_open_lock );return (0 ); }static int nvram_release (struct inode *inode,struct file *file ) { -lock_kernel (); +spin_lock ( &nvram_open_lock ); nvram_open_cnt--;if (file->f_flags &O_EXCL ) nvram_open_mode &= ~NVRAM_EXCL ;if (file->f_mode &2 ) nvram_open_mode &= ~NVRAM_WRITE ; -unlock_kernel (); - +spin_unlock ( &nvram_open_lock );return (0 ); }
drivers/char/pc110pad.c +12 -5
@@ -68,7 +68,9 @@/* driver/filesystem interface management */ static wait_queue_head_t queue;static struct fasync_struct *asyncptr; -static int active;/* number of concurrent open()s */ +static active_count = 0; /* number of concurrent open()s */ +static spinlock_t active_lock =SPIN_LOCK_UNLOCKED ; +/* this lock should be held when referencing active_count */ static struct semaphore reader_lock;/**
@@ -584,11 +586,11 @@static int close_pad (struct inode * inode,struct file * file) { -lock_kernel ();fasync_pad (-1 , file,0 ); -if (!--active) +spin_lock ( &active_lock ); +if (!--active_count)outb (0x30 , current_params.io+2 );/* switch off digitiser */ -unlock_kernel (); +spin_unlock ( &active_lock );return 0 ; }
@@ -608,8 +610,13 @@ {unsigned long flags; -if (active++) +spin_lock ( &active_lock ); +if (active_count++) + { +spin_unlock ( &active_lock );return 0 ; + } +spin_unlock ( &active_lock );save_flags (flags);cli ();
drivers/char/pc_keyb.c +21 -10
@@ -90,6 +90,7 @@static struct aux_queue *queue;/* Mouse data buffer. */ static int aux_count; +static spinlock_t aux_count_lock =SPIN_LOCK_UNLOCKED ;/* used when we send commands to the mouse that expect an ACK. */ static unsigned char mouse_reply_expected;
@@ -405,8 +406,9 @@if (rqst ==PM_RESUME ) {if (queue) {/* Aux port detected */ -if (aux_count ==0 ) {/* Mouse not in use */ -spin_lock_irqsave (&kbd_controller_lock, flags); +spin_lock_irqsave ( &aux_count_lock, flags); +if ( aux_count ==0 ) {/* Mouse not in use */ +spin_lock (&kbd_controller_lock);/* * Dell Lat. C600 A06 enables mouse after resume. * When user touches the pad, it posts IRQ 12
@@ -418,8 +420,9 @@kbd_write_command (KBD_CCMD_WRITE_MODE );kb_wait ();kbd_write_output (AUX_INTS_OFF ); - spin_unlock_irqrestore(&kbd_controller_lock, flags); + spin_unlock(&kbd_controller_lock, flags); } +spin_unlock_irqrestore ( &aux_count_lock,flags ); } }#endif
@@ -430,6 +433,7 @@static inline void handle_mouse_event (unsigned char scancode) {#ifdef CONFIG_PSMOUSE +int flags;static unsigned char prev_code;if (mouse_reply_expected) {if (scancode ==AUX_ACK ) {
@@ -448,7 +452,8 @@ prev_code = scancode;add_mouse_randomness (scancode); -if (aux_count) { +spin_lock_irqsave ( &aux_count_lock, flags); +if ( aux_count ) {int head = queue->head; queue->buf[head] = scancode;
@@ -459,6 +464,7 @@ wake_up_interruptible(&queue->proc_list); } } +spin_unlock_irqrestore ( &aux_count_lock, flags);#endif }
@@ -1046,16 +1052,17 @@static int release_aux (struct inode * inode,struct file * file) { -lock_kernel (); +int flags;fasync_aux (-1 , file,0 ); -if (--aux_count) { -unlock_kernel (); +spin_lock_irqsave ( &aux_count, flags ); +if ( --aux_count ) { +spin_unlock_irqrestore ( &aux_count_lock );return 0 ; } +spin_unlock_irqrestore ( &aux_count_lock, flags );kbd_write_cmd (AUX_INTS_OFF );/* Disable controller ints */ kbd_write_command_w (KBD_CCMD_MOUSE_DISABLE );aux_free_irq (AUX_DEV ); -unlock_kernel ();return 0 ; }
@@ -1066,12 +1073,16 @@static int open_aux (struct inode * inode,struct file * file) { -if (aux_count++) { +int flags; +spin_lock_irqsave ( &aux_count_lock, flags ); +if ( aux_count++ ) { +spin_unlock_irqrestore ( &aux_count_lock );return 0 ; } queue->head = queue->tail =0 ;/* Flush input queue */ if (aux_request_irq (keyboard_interrupt,AUX_DEV )) { aux_count--; +spin_unlock_irqrestore ( &aux_count_lock, flags );return -EBUSY ; }kbd_write_command_w (KBD_CCMD_MOUSE_ENABLE );/* Enable the
@@ -1083,7 +1094,7 @@mdelay (2 );/* Ensure we follow the kbc access delay rules.. */ send_data (KBD_CMD_ENABLE );/* try to workaround toshiba4030cdt problem */ - +spin_unlock_irqrestore ( &aux_count_lock, flags );return 0 ; }
drivers/char/pcwd.c +9 -8
@@ -100,7 +100,8 @@#define WD_SRLY2 0x80/* Software external relay triggered */ static int current_readport, revision, temp_panic; -static int is_open, initial_status, supports_temp, mode_debug; +static atomic_t open_allowed =ATOMIC_INIT (1 ); +static int initial_status, supports_temp, mode_debug;static spinlock_t io_lock;/*
@@ -402,9 +403,12 @@switch (MINOR (ino->i_rdev)) {case WATCHDOG_MINOR : -if (is_open) +if ( !atomic_dec_and_test (&open_allowed) ) + { +atomic_inc ( &open_allowed );return -EBUSY ; -MOD_INC_USE_COUNT ; + } +MOD_INC_USE_COUNT ;/* Enable the port */ if (revision ==PCWD_REVISION_C ) {
@@ -412,7 +416,6 @@outb_p (0x00 , current_readport +3 );spin_unlock (&io_lock); } - is_open =1 ;return (0 );case TEMP_MINOR :return (0 );
@@ -452,8 +455,6 @@ {if (MINOR (ino->i_rdev)==WATCHDOG_MINOR ) { -lock_kernel (); - is_open =0 ;#ifndef CONFIG_WATCHDOG_NOWAYOUT /* Disable the board */ if (revision ==PCWD_REVISION_C ) {
@@ -462,8 +463,8 @@outb_p (0xA5 , current_readport +3 ); spin_unlock(&io_lock); } +atomic_inc ( &open_allowed );#endif -unlock_kernel (); }return 0 ; }
@@ -574,7 +575,7 @@printk ("pcwd: v%s Ken Hollis ([email protected])\n" ,WD_VER );/* Initial variables */ - is_open =0 ; +set_bit (0 , &open_allowed ); supports_temp =0 ; mode_debug =0 ; temp_panic =0 ;
drivers/char/ppdev.c +0 -2
@@ -646,7 +646,6 @@struct pp_struct *pp = file->private_data;int compat_negot; -lock_kernel (); compat_negot =0 ;if (!(pp->flags &PP_CLAIMED ) && pp->pdev && (pp->state.mode !=IEEE1284_MODE_COMPAT )) {
@@ -695,7 +694,6 @@ printk (KERN_DEBUG CHRDEV "%x: unregistered pardevice\n", minor); } -unlock_kernel ();kfree (pp);
drivers/char/qpmouse.c +8 -2
@@ -111,6 +111,7 @@static int qp_present;static int qp_count; +static spinlock_t qp_count_lock =SPIN_LOCK_UNLOCKED ;static int qp_data =QP_DATA ;static int qp_status =QP_STATUS ;
@@ -141,8 +142,8 @@ {unsigned char status; -lock_kernel ();fasync_qp (-1 , file,0 ); +spin_lock ( &qp_count_lock );if (!--qp_count) {if (!poll_qp_status ())printk (KERN_WARNING "Warning: Mouse device busy in release_qp()\n" );
@@ -152,7 +153,7 @@printk (KERN_WARNING "Warning: Mouse device busy in release_qp()\n" );free_irq (QP_IRQ ,NULL ); } -unlock_kernel (); +spin_unlock ( &qp_count_lock );return 0 ; }
@@ -168,8 +169,13 @@if (!qp_present)return -EINVAL ; +spin_lock ( &qp_count_lock );if (qp_count++) + { +spin_unlock ( &qp_count_lock );return 0 ; + } +spin_unlock ( &qp_count_lock );if (request_irq (QP_IRQ , qp_interrupt,0 ,"PS/2 Mouse" ,NULL )) { qp_count--;
drivers/char/qtronix.c +0 -2
@@ -492,10 +492,8 @@static int release_aux (struct inode * inode,struct file * file) { -lock_kernel ();fasync_aux (-1 , file,0 ); aux_count--; -unlock_kernel ();return 0 ; }
drivers/char/sbc60xxwdt.c +0 -2
@@ -214,7 +214,6 @@static int fop_close (struct inode * inode,struct file * file) { -lock_kernel ();if (MINOR (inode->i_rdev) ==WATCHDOG_MINOR ) {if (wdt_expect_close)
@@ -225,7 +224,6 @@ } } wdt_is_open =0 ; -unlock_kernel ();return 0 ; }
drivers/char/softdog.c +0 -2
@@ -100,12 +100,10 @@* Shut off the timer. * Lock it in if it's a module and we defined ...NOWAYOUT */ -lock_kernel ();#ifndef CONFIG_WATCHDOG_NOWAYOUT del_timer (&watchdog_ticktock);#endif timer_alive=0 ; -unlock_kernel ();return 0 ; }
drivers/char/tpqic02.c +0 -2
@@ -2395,7 +2395,6 @@ {kdev_t dev = inode->i_rdev; -lock_kernel ();if (TP_DIAGS (dev)) {printk ("qic02_tape_release: dev=%s\n" ,kdevname (dev)); }
@@ -2419,7 +2418,6 @@ (void )do_qic_cmd (QCMD_REWIND ,TIM_R ); } } -unlock_kernel ();return 0 ; }/* qic02_tape_release */
drivers/char/w83877f_wdt.c +0 -2
@@ -214,7 +214,6 @@static int fop_close (struct inode * inode,struct file * file) { -lock_kernel ();if (MINOR (inode->i_rdev) ==WATCHDOG_MINOR ) {if (wdt_expect_close)
@@ -225,7 +224,6 @@ } } wdt_is_open =0 ; -unlock_kernel ();return 0 ; }
drivers/char/wdt.c +0 -2
@@ -374,7 +374,6 @@static int wdt_release (struct inode *inode,struct file *file) { -lock_kernel ();if (MINOR (inode->i_rdev)==WATCHDOG_MINOR ) {#ifndef CONFIG_WATCHDOG_NOWAYOUT
@@ -383,7 +382,6 @@#endif wdt_is_open=0 ; } -unlock_kernel ();return 0 ; }
drivers/char/wdt285.c +0 -2
@@ -94,10 +94,8 @@static int watchdog_release (struct inode *inode,struct file *file) {#ifdef ONLY_TESTING -lock_kernel ();free_irq (IRQ_TIMER4 ,NULL ); timer_alive =0 ; -unlock_kernel ();#else /* * It's irreversible!
drivers/char/wdt977.c +0 -2
@@ -92,7 +92,6 @@* Lock it in if it's a module and we defined ...NOWAYOUT */ #ifndef CONFIG_WATCHDOG_NOWAYOUT -lock_kernel ();// unlock the SuperIO chip outb (0x87 ,0x370 );
@@ -124,7 +123,6 @@outb (0xAA ,0x370 ); timer_alive=0 ; -unlock_kernel ();printk (KERN_INFO "Watchdog: shutdown.\n" );#endif
drivers/char/wdt_pci.c +4 -6
@@ -353,16 +353,16 @@switch (MINOR (inode->i_rdev)) {case WATCHDOG_MINOR : -if (wdt_is_open) +if (test_and_set_bit (0 ,&wdt_is_open) ) + {return -EBUSY ; + }#ifdef CONFIG_WATCHDOG_NOWAYOUT MOD_INC_USE_COUNT ;#endif /* * Activate */ - - wdt_is_open=1 ;inb_p (WDT_DC );/* Disable */
@@ -412,13 +412,11 @@ {if (MINOR (inode->i_rdev)==WATCHDOG_MINOR ) { -lock_kernel ();#ifndef CONFIG_WATCHDOG_NOWAYOUT inb_p (WDT_DC );/* Disable counters */ wdtpci_ctr_load (2 ,0 );/* 0 length reset pulses now */ #endif - wdt_is_open=0 ; -unlock_kernel (); +clear_bit (0 , &wdt_is_open ); }return 0 ; }
drivers/i2c/i2c-dev.c +0 -5
@@ -423,14 +423,9 @@#endif #if LINUX_KERNEL_VERSION <KERNEL_VERSION (2 ,4 ,0 )MOD_DEC_USE_COUNT ; -#else /* LINUX_KERNEL_VERSION >= KERNEL_VERSION(2,4,0) */ -lock_kernel ();#endif /* LINUX_KERNEL_VERSION < KERNEL_VERSION(2,4,0) */ if (i2cdev_adaps[minor]->dec_use) i2cdev_adaps[minor]->dec_use (i2cdev_adaps[minor]); -#if LINUX_KERNEL_VERSION >=KERNEL_VERSION (2 ,4 ,0 ) -unlock_kernel (); -#endif /* LINUX_KERNEL_VERSION >= KERNEL_VERSION(2,4,0) */ return 0 ; }
drivers/ide/ide-tape.c +0 -2
@@ -5486,7 +5486,6 @@idetape_pc_t pc;unsigned int minor=MINOR (inode->i_rdev); -lock_kernel (); tape = drive->driver_data;#if IDETAPE_DEBUG_LOG if (tape->debug_level >=3 )
@@ -5517,7 +5516,6 @@MOD_DEC_USE_COUNT ; }clear_bit (IDETAPE_BUSY , &tape->flags); -unlock_kernel ();return 0 ; }
Only in linux/drivers/ide: ide-tape.c.orig
drivers/ieee1394/raw1394.c +0 -2
@@ -949,7 +949,6 @@struct pending_request *req;int done =0 , i; -lock_kernel ();for (i =0 ; i <64 ; i++) {if (fi->listen_channels & (1ULL << i)) {hpsb_unlisten_channel (hl_handle, fi->host, i);
@@ -990,7 +989,6 @@kfree (fi);V22_COMPAT_MOD_DEC_USE_COUNT ; -unlock_kernel ();return 0 ; }
drivers/input/evdev.c +0 -2
@@ -94,7 +94,6 @@struct evdev_list *list = file->private_data;struct evdev_list **listptr; -lock_kernel (); listptr = &list->evdev->list;evdev_fasync (-1 , file,0 );
@@ -113,7 +112,6 @@ }kfree (list); -unlock_kernel ();return 0 ; }
drivers/input/input.c +0 -2
@@ -387,9 +387,7 @@ old_fops = file->f_op; file->f_op = new_fops; -lock_kernel (); err = new_fops->open (inode, file); -unlock_kernel ();if (err) {fops_put (file->f_op);
drivers/input/joydev.c +1 -3
@@ -164,8 +164,7 @@ {struct joydev_list *list = file->private_data;struct joydev_list **listptr; - -lock_kernel (); + listptr = &list->joydev->list;joydev_fasync (-1 , file,0 );
@@ -184,7 +183,6 @@ }kfree (list); -unlock_kernel ();return 0 ; }
drivers/input/mousedev.c +1 -3
@@ -169,8 +169,7 @@ {struct mousedev_list *list = file->private_data;struct mousedev_list **listptr; - -lock_kernel (); + listptr = &list->mousedev->list;mousedev_fasync (-1 , file,0 );
@@ -208,7 +207,6 @@ }kfree (list); -unlock_kernel ();return 0 ; }
drivers/isdn/avmb1/capi.c +0 -4
@@ -1055,7 +1055,6 @@ {struct capidev *cdev = (struct capidev *)file->private_data; -lock_kernel ();capincci_free (cdev,0xffffffff );capidev_free (cdev); file->private_data =NULL ;
@@ -1064,7 +1063,6 @@#ifdef _DEBUG_REFCOUNTprintk (KERN_DEBUG "capi_release %d\n" ,GET_USE_COUNT (THIS_MODULE ));#endif -unlock_kernel ();return 0 ; }
@@ -1243,13 +1241,11 @@struct capiminor *mp = (struct capiminor *)file->private_data;if (mp) { -lock_kernel (); mp->file =0 ;if (mp->nccip ==0 ) {capiminor_free (mp); file->private_data =NULL ; } -unlock_kernel (); }#ifdef _DEBUG_REFCOUNT
drivers/isdn/divert/divert_procfs.c +8 -12
@@ -29,6 +29,7 @@ulong if_used =0 ;/* number of interface users */ static struct divert_info *divert_info_head =NULL ;/* head of queue */ static struct divert_info *divert_info_tail =NULL ;/* pointer to last entry */ +static spinlock_t divert_info_lock =SPIN_LOCK_UNLOCKED ;/* lock for queue */ static wait_queue_head_t rd_queue;/*********************************/
@@ -50,8 +51,7 @@return ;/* no memory */ strcpy (ib->info_start, cp);/* set output string */ ib->next =NULL ; -save_flags (flags); -cli (); +spin_lock_irqsave ( &divert_info_lock, flags ); ib->usage_cnt = if_used;if (!divert_info_head) divert_info_head = ib;/* new head */
@@ -70,6 +70,7 @@ }else break ; }/* divert_info_head->next */ +spin_lock_irqrestore ( &divert_info_lock, flags );wake_up_interruptible (&(rd_queue)); }/* put_info_buffer */
@@ -135,17 +136,14 @@ {unsigned long flags; -lock_kernel (); -save_flags (flags); -cli (); - if_used++; +spin_lock_irqsave ( &divert_info_lock, flags ); + if_used++;if (divert_info_head) (struct divert_info **) filep->private_data = &(divert_info_tail->next);else (struct divert_info **) filep->private_data = &divert_info_head; -restore_flags (flags); +spin_unlock_irqrestore ( &divert_info_lock, flags );/* start_divert(); */ -unlock_kernel ();return (0 ); }/* isdn_divert_open */
@@ -158,9 +156,7 @@struct divert_info *inf;unsigned long flags; -lock_kernel (); -save_flags (flags); -cli (); +spin_lock_irqsave ( &divert_info_lock, flags ); if_used--; inf = *((struct divert_info **) filep->private_data);while (inf) {
@@ -174,7 +170,7 @@ divert_info_head = divert_info_head->next;kfree (inf); } -unlock_kernel (); +spin_unlock_irq ( &divert_info_lock, flags );return (0 ); }/* isdn_divert_close */
drivers/macintosh/adb.c +0 -2
@@ -672,7 +672,6 @@struct adbdev_state *state = file->private_data;unsigned long flags; -lock_kernel ();if (state) { file->private_data =NULL ;spin_lock_irqsave (&state->lock, flags);
@@ -685,7 +684,6 @@spin_unlock_irqrestore (&state->lock, flags); } } -unlock_kernel ();return 0 ; }
drivers/macintosh/via-pmu.c +0 -2
@@ -1983,7 +1983,6 @@struct pmu_private *pp = file->private_data;unsigned long flags; -lock_kernel ();if (pp !=0 ) { file->private_data =0 ;spin_lock_irqsave (&all_pvt_lock, flags);
@@ -1991,7 +1990,6 @@spin_unlock_irqrestore (&all_pvt_lock, flags);kfree (pp); } -unlock_kernel ();return 0 ; }
drivers/message/i2o/i2o_config.c +0 -2
@@ -847,7 +847,6 @@struct i2o_cfg_info *p1, *p2;unsigned long flags; -lock_kernel (); p1 = p2 =NULL ;spin_lock_irqsave (&i2o_config_lock, flags);
@@ -870,7 +869,6 @@ p1 = p1->next; }spin_unlock_irqrestore (&i2o_config_lock, flags); -unlock_kernel ();return 0 ; }
drivers/net/ppp_generic.c +0 -2
@@ -325,7 +325,6 @@ {struct ppp_file *pf = (struct ppp_file *) file->private_data; -lock_kernel ();if (pf !=0 ) { file->private_data =0 ;if (atomic_dec_and_test (&pf->refcnt)) {
@@ -339,7 +338,6 @@ } } } -unlock_kernel ();return 0 ; }
drivers/net/wan/cosa.c +0 -2
@@ -974,13 +974,11 @@struct cosa_data *cosa;unsigned long flags; -lock_kernel (); cosa = channel->cosa;spin_lock_irqsave (&cosa->lock, flags); cosa->usage--; channel->usage--;spin_unlock_irqrestore (&cosa->lock, flags); -unlock_kernel ();return 0 ; }
drivers/pcmcia/ds.c +0 -2
@@ -597,7 +597,6 @@DEBUG (0 ,"ds_release(socket %d)\n" , i);if ((i >= sockets) || (sockets ==0 ))return 0 ; -lock_kernel (); s = &socket_table[i]; user = file->private_data;if (CHECK_USER (user))
@@ -615,7 +614,6 @@ user->user_magic =0 ;kfree (user); out: -unlock_kernel ();return 0 ; }/* ds_release */
drivers/pnp/isapnp_proc.c +0 -4
@@ -170,12 +170,10 @@kfree (buffer);return -ENOMEM ; } -lock_kernel (); buffer->curr = buffer->buffer; file->private_data = buffer;if (mode ==O_RDONLY )isapnp_info_read (buffer); -unlock_kernel ();return 0 ; }
@@ -187,12 +185,10 @@if ((buffer = (isapnp_info_buffer_t *) file->private_data) ==NULL )return -EINVAL ; mode = file->f_flags &O_ACCMODE ; -lock_kernel ();if (mode ==O_WRONLY )isapnp_info_write (buffer);vfree (buffer->buffer);kfree (buffer); -unlock_kernel ();return 0 ; }
drivers/scsi/sg.c +0 -3
@@ -336,9 +336,7 @@Sg_device * sdp;Sg_fd * sfp; -lock_kernel ();if ((! (sfp = (Sg_fd *)filp->private_data)) || (! (sdp = sfp->parentdp))) { -unlock_kernel ();return -ENXIO ; }SCSI_LOG_TIMEOUT (3 ,printk ("sg_release: dev=%d\n" ,MINOR (sdp->i_rdev)));
@@ -352,7 +350,6 @@ sdp->exclude =0 ;wake_up_interruptible (&sdp->o_excl_wait); } -unlock_kernel ();return 0 ; }
drivers/sgi/char/graphics.c +0 -2
@@ -196,7 +196,6 @@int board =GRAPHICS_CARD (inode->i_rdev);/* Tell the rendering manager that one client is going away */ -lock_kernel ();rrm_close (inode, file);/* Was this file handle from the board owner?, clear it */
@@ -206,7 +205,6 @@ (*cards [board].g_reset_console)();enable_gconsole (); } -unlock_kernel ();return 0 ; }
drivers/sgi/char/shmiq.c +19 -14
@@ -79,11 +79,13 @@/* /dev/qcntlN attached memory regions, location and size of the event queue */ static struct { - int opened; /* if this device has been opened */ - void *shmiq_vaddr; /* mapping in kernel-land */ - int tail; /* our copy of the shmiq->tail */ - int events; - int mapped; + int opened; + void *shmiq_vaddr; /* mapping in kernel-land */ + spinlock_t shmiq_lock:SPIN_LOCK_UNLOCKED; + /* protects vaddr and opened */ + int tail; /* our copy of the shmiq->tail */ + int events; + int mapped; wait_queue_head_t proc_list; struct fasync_struct *fasync;
@@ -328,10 +330,10 @@ size = vma->vm_end - vma->vm_start; start = vma->vm_start; -lock_kernel (); +spin_lock ( &shmiqs [minor].shmiq_lock ); mem = (unsigned long ) shmiqs [minor].shmiq_vaddr =vmalloc_uncached (size);if (!mem) { -unlock_kernel (); +spin_unlock ( &shmiqs [minor].shmiq_lock );return -EINVAL ; }
@@ -347,8 +349,7 @@ shmiqs [minor].tail =0 ;/* Init the shared memory input queue */ memset (shmiqs [minor].shmiq_vaddr,0 , size); -unlock_kernel (); - +spin_unlock ( &shmiqs [minor].shmiq_lock );return error; }
@@ -393,13 +394,16 @@ minor--;if (minor >MAX_SHMI_QUEUES )return -EINVAL ; +spin_lock ( &shmiqs [minor].shmiq_lock );if (shmiqs [minor].opened) + { +spin_unlock ( &shmiqs [minor].shmiq_lock );return -EBUSY ; + } -lock_kernel (); shmiqs [minor].opened =1 ; shmiqs [minor].shmiq_vaddr =0 ; -unlock_kernel (); +spin_unlock ( &shmiqs [minor].shmiq_lock );return 0 ; }
@@ -429,18 +433,19 @@if (minor >MAX_SHMI_QUEUES )return -EINVAL ; + +spin_lock ( &shmiqs [minor].shmiq_lock );if (shmiqs [minor].opened ==0 )return -EINVAL ; -lock_kernel ();shmiq_qcntl_fasync (-1 , filp,0 ); - shmiqs [minor].opened =0 ; + shmiqs [minor].opened =0 ; shmiqs [minor].mapped =0 ; shmiqs [minor].events =0 ; shmiqs [minor].fasync =0 ;vfree (shmiqs [minor].shmiq_vaddr); shmiqs [minor].shmiq_vaddr =0 ; -unlock_kernel (); +spin_unlock ( &shmiqs [minor].shmiq_lock );return 0 ; }
drivers/sgi/char/streamable.c +0 -2
@@ -221,9 +221,7 @@static int sgi_mouse_close (struct inode *inode,struct file *filp) { -lock_kernel (); mouse_opened =0 ; -unlock_kernel ();return 0 ; }
drivers/usb/dabusb.c +0 -2
@@ -622,7 +622,6 @@dbg ("dabusb_release" ); -lock_kernel ();down (&s->mutex);dabusb_stop (s);dabusb_free_buffers (s);
@@ -636,7 +635,6 @@wake_up (&s->remove_ok); s->opened =0 ; -unlock_kernel ();return 0 ; }
drivers/usb/hiddev.c +0 -2
@@ -193,7 +193,6 @@struct hiddev_list *list = file->private_data;struct hiddev_list **listptr; -lock_kernel (); listptr = &list->hiddev->list;hiddev_fasync (-1 , file,0 );
@@ -209,7 +208,6 @@ }kfree (list); -unlock_kernel ();return 0 ; }
fs/autofs4/root.c +0 -4
@@ -202,8 +202,6 @@ {struct autofs_info *inf; -lock_kernel (); -DPRINTK (("autofs4_dentry_release: releasing %p\n" , de)); inf =autofs4_dentry_ino (de);
@@ -215,8 +213,6 @@autofs4_free_ino (inf); } - -unlock_kernel (); }/* For dentries of directories in the root dir */
Quoted 2 lines
drivers' release functions. The release functions are already serialized in the VFS code by an
atomic_twhich guarantees that each
The release function was only partially serialized - what stops a release and an open racing each other ? That in several cases was why the lock was there.
I'm not saying your patch isnt a good thing. General rule of thumb, any user of lock_kernel has a race 8)
Quoted 6 lines · Alan Cox
Quoted 2 lines
drivers' release functions. The release functions are already serialized in the VFS code by an
atomic_twhich guarantees that eachThe release function was only partially serialized - what stops a release and an open racing each other ? That in several cases was why the lock was there.
Nothing, because the BKL is not held for all opens anymore. In most of the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or used atomic operations to keep open from racing with release.
Quoted 5 lines · David C. Hansen
Quoted 3 lines · Alan Cox
The release function was only partially serialized - what stops a release and an open racing each other ? That in several cases was why the lock was there.
->release() is not serialized AT ALL. It is serialized for given struct file, but call open(2) twice and you've got two struct file for the same device. close() both and you've got two calls of ->release(), quite possibly - simultaneous.
Quoted 1 line
Nothing, because the BKL is not held for all opens anymore. In most of
RTFS. fs/devices.c::chrdev_open().
Quoted 2 lines
the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or
See above.
Quoted 1 line
used atomic operations to keep open from racing with release.
Right. Including quite a few where you schedule under that spinlock.
In other words, patch is completely bogus. BKL removal may be a good idea, but you really need to audit the code. Which requires at least some understanding of the things you are doing. There _are_ races and they need to be dealt with. But blind BKL removal doesn't fix any and breaks quite a few places where the code was actually correct.
Quoted 4 lines · Alexander Viro
->release() is not serialized AT ALL. It is serialized for given struct file, but call open(2) twice and you've got two struct file for the same device.
close()both and you've got two calls of ->release(), quite possibly - simultaneous.
OK, that clears some things up. So, the file->fcount is only used in cases where the file descriptor was dup'd, right?
As Rick Lindsley pointed out to me:
Quoted 7 lines
In cases where we removed BKL from
release()and left obvious locking issues as "an exercise to the reader", we MAY have broken things because the BKL (we now know) may have been serializing opens and closes.
In cases where we replaced it with atomic locking or a spinlock, we've done nothing but replace one lock with another (unless there are subtleties
back to Alexander Viro:
Quoted 1 line
In other words, patch is completely bogus.
No, not completely. In a lot of cases we just replaced some regular arithmetic with atomic instructions of some sort. These changes are still completely valid. But, in the cases where we added locking, we need to reevaluate them for potential problem. In the cases where we just removed the BKL, we really need to check them to make sure that we didn't introduce anything. Thanks for the feedback, Al! This has been very helpful!
--
Dave Hansen
[email protected]
Quoted 7 lines · David C. Hansen
Quoted 4 lines · Alexander Viro
->release() is not serialized AT ALL. It is serialized for given struct file, but call open(2) twice and you've got two struct file for the same device.
close()both and you've got two calls of ->release(), quite possibly - simultaneous.OK, that clears some things up. So, the
file->fcountis only used in cases where the file descriptor was dup'd, right?
Think for a minute - current IO position is in file->f_pos. So if two descriptors have independent current positions (lseek() on one of them doesn't affect another) they have to have different instances of struct file behind them.
struct file is an opened channel
descriptor refers to opened channelfile->f_count is the number of references to _this_ channel dup() creates a descriptor refering to the same channel so does fork()open() creates a new channel and descriptor refering to it.
That difference between descriptors and opened channels exists in any UNIX - otherwise you would either have lseek() done in one program screwing everybody else or (if fork would create new channels instead of new references to channels) have (ls; ls) > foo break (shell opens foo and both ls(1) inherit it over fork(); you _really_ want changes in current positions resulting from the first ls to affect the stdout of the second one).
Quoted 5 lines
back to Alexander Viro:
Quoted 1 line
In other words, patch is completely bogus.
No, not completely. In a lot of cases we just replaced some regular arithmetic with atomic instructions of some sort. These changes are still completely valid. But, in the cases where we added locking, we
Valid, but not necessary a good idea. Notice that there are very real races in ->release() and ->open() instances. Removing BKL from these (or replacing it with atomic operations) doesn't fix the existing race. Moreover, it creates a false impression that thing had been reviewed and fixed.
Quoted 3 lines
need to reevaluate them for potential problem. In the cases where we just removed the BKL, we really need to check them to make sure that we didn't introduce anything.
You really need that in all cases. Sorry for "told you so", but that's what I'd been trying to explain from the very beginning - back when the idea of that patch had been discussed the first time. That work really has to be done driver-by-driver...
Quoted 5 lines
back to Alexander Viro:
Quoted 1 line
In other words, patch is completely bogus.
No, not completely. In a lot of cases we just replaced some regular arithmetic with atomic instructions of some sort. These changes are still completely valid. But, in the cases where we added locking, we
Valid, but not necessary a good idea. Notice that there are very real races in ->release() and ->open() instances. Removing BKL from these (or replacing it with atomic operations) doesn't fix the existing race. Moreover, it creates a false impression that thing had been reviewed and fixed.
Well, but for the revelation that the BKL is held in chrdev_open(), offering implicit serialization for all open routines, these *were* reviewed and thought fixed. Atomic operations *do* fill the bill when we are looking at counts used to guarantee exclusive opens. They generally *don't* fill the bill when you want to take virtually any other action based on those counts.
Sorry for "told you so", but that's what I'd been trying to explain from the very beginning - back when the idea of that patch had been discussed the first time. That work really has to be done driver-by-driver...
It actually *was* in this case. This may look like an afternoon of editor exercise, but each of the cases was inspected and reviewed before recommending the patch, and collectively represents several weeks' worth of work. In early November I posted a note to linux-kernel which said, in part:
Before we post these patches, I thought I'd ask if in the time since Al Viro moved this out here (July 2000) if anybody (especially him!) has found a *legitimate* use of the BKL in the release() functions. (We have not found one.)
There was no response.
Given the miss on chrdev_open(), we've got some patch patches to come up with. However, this isn't a dump-and-run job. We accept the responsibility for those patches, and we'll patch 'em up. While we can't then miraculously announce it's safe to remove lock_kernel around opens, we CAN announce we are closer, and nothing else will be any more broken than it was before we started. As I said in an earlier post (and I don't think anybody disagrees): this is an incremental task. Any assistance you can provide in reviewing this work, including all to date, is welcome.
Rick
Quoted 3 lines
opens, we CAN announce we are closer, and nothing else will be any more broken than it was before we started. As I said in an earlier post (and I don't think anybody disagrees): this is an incremental task. Any
This is why we have a development tree. Its moving things in the right direction which is important. I suspect many drivers will want to use semaphores rather than atomic counts however, to ensure that an open doesn't complete while a previous release is still shutting down hardware
Alan
This is why we have a development tree. Its moving things in the right direction which is important. I suspect many drivers will want to use semaphores rather than atomic counts however, to ensure that an open doesn't complete while a previous release is still shutting down hardware
Yes, the only successful application for atomic counts that I've seen (in this context) is for exclusive open code that looks like
if (count++) {
count--;
return -EBUSY;
}in the open routine and
count--;
in the release. If you want to do anything else as a result of that count, you'll need additional locking because it could have changed a nanosecond after it was (safely) incremented or decremented. You can't count on it remaining that value after you check it.
release()s that want to shutdown the device, free memory, or take other actions will want to employ either a spinlock (plain or r/w as appropriate) or a sleeping semaphore to insure things remain stable until those actions are complete.
Rick
I think you're exactly the right person to be helping out with this; glad to see you join the discussion.
The biggest flaw in the patch was not realizing that chrdev_open held the BKL for us during opens as well. I don't think that invalidates it in its entirety though. In cases where the patch replaced BKL with some other form of locking, we should be no worse off (but for your concerns about scheduling) because we added it in the open routines too. In cases where we removed BKL from release but noted there were "still" locking issues, yes, we need to examine those closely and fix them in case we've created a locking issue where none existed before.
In most (all?) cases, the spinlocks are held briefly and (we believe) not across scheduling opportunities. If you see areas of the patch where that is not true, I agree they should be addressed and please point them out.
Rick
Quoted 5 lines · Alexander Viro
In other words, patch is completely bogus. BKL removal may be a good idea, but you really need to audit the code. Which requires at least some understanding of the things you are doing. There _are_ races and they need to be dealt with. But blind BKL removal doesn't fix any and breaks quite a few places where the code was actually correct.
People seem to believe that BKL is worse than added spin-locks, but I very much doubt this is always true.
Quoted 6 lines
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
People seem to believe that BKL is worse than added spin-locks, but I very much doubt this is always true.
Depends, to some extent, on your definition of "worse". To address
two: the BKL (in general) encourages "worse" design and promotes "worse" performance. Combine these with a general lack of comments where clever and subtle code is being employed and the pervasive use of the BKL can make debugging (or design change) difficult.
The BKL *is* a spinlock, with some questionable special handling in other parts of the code which permits and encourages poor programming (like automatically releasing it before you go to sleep, and allowing you to "acquire" it multiple times). In general, if you do not need the special handling, there are compelling reasons to replace it with a purpose-specific lock (or locking mechanism).
Setting aside architectural and academic issues of "locking design", the main thing the current practice does, from a performance standpoint, is create a generic bottleneck. "One lock to rule them all ..." It creates strange effects like heavy activity in one subsystem suddenly bogging down a completely disjoint subsystem. Two separate locks, in that case, would allow one subsystem to be very busy without affecting the other.
The advantages gained become more evident on multiple-cpu machines. Since this is not your most common desktop variant :) the trick is to introduce changes which allow that hardware to excel while not impacting the more common case: the single cpu desktop or laptop. This is sometimes easy; often, however, it is not, and so the work goes slowly.
Locks should have clearly defined purpose and scope, just like any other component of a software system. The BKL has evolved/mutated beyond its original scope and needs to be trimmed back. When it is trimmed back to either one or zero purposes then, I believe, we can put down the weed-whackers and stop trimming.
Rick
Quoted 4 lines · David C. Hansen
Nothing, because the BKL is not held for all opens anymore. In most of the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or used atomic operations to keep open from racing with release.
All char and block devs are opened with the BKL held - see chrdev_open in fs/devices.c and do_open in fs/block_dev.c
--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html
Quoted 8 lines · Russell King
Quoted 2 lines · David C. Hansen
Nothing, because the BKL is not held for all opens anymore. In most of the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or used atomic operations to keep open from racing with release.
All char and block devs are opened with the BKL held - see
chrdev_openinfs/devices.canddo_openinfs/block_dev.c
I wrote a quick and dirty char device driver to see if this happened. If I run two tasks doing a bunch of opens and closes, the -EBUSY condition in the open function does happen. Is my driver doing something wrong?
Here is the meat of the driver:
static int Device_Open = 0;
int testdev_open(struct inode *inode, struct file *file) {
if ( test_and_set_bit(0,&Device_Open) ) {
printk( "attempt to open testdev more than once\n" );
return -EBUSY; }
MOD_INC_USE_COUNT;
return SUCCESS;
}
int testdev_release(struct inode *inode, struct file *file) {
clear_bit(0,&Device_Open);
MOD_DEC_USE_COUNT;
return 0;
}
static int Major;
struct file_operations Fops = {
open: testdev_open,
release: testdev_release,
};
/* Initialize the module - Register the character device */ int init_module(void)
{
Major = register_chrdev(0,
DEVICE_NAME,
&Fops); /* Negative values signify an error */
if (Major < 0) {
printk ("%s: %s device failed with %d\n",MODULE_NAME,
"Sorry, registering the character",
Major);
return Major;
}
printk( "%s: loaded successfully on Major:%d\n",MODULE_NAME,Major); return 0;
}
void cleanup_module(void)
{
int ret;
ret = unregister_chrdev(Major, DEVICE_NAME); if (ret < 0)
printk("%s: Error in unregister_MODULE_NAME: %d\n",
MODULE_NAME, ret); }
Quoted 33 lines · David C. Hansen
Quoted 11 lines · Russell King
Quoted 5 lines · David C. Hansen
Nothing, because the BKL is not held for all opens anymore. In most of the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or used atomic operations to keep open from racing with release.
All char and block devs are opened with the BKL held - see
chrdev_openinfs/devices.canddo_openinfs/block_dev.cI wrote a quick and dirty char device driver to see if this happened.
If I run two tasks doing a bunch of opens and closes, the-EBUSY
condition in the open function does happen. Is my driver doing something wrong?Here is the meat of the driver:
static int
Device_Open= 0;int
testdev_open(struct inode *inode, struct file *file){if ( test_and_set_bit(0,&Device_Open) ) { printk( "attempt to open testdev more than once\n" ); return -EBUSY; }
MOD_INC_USE_COUNT;
return SUCCESS;
}
Quoted 6 lines
int
testdev_release(struct inode *inode, struct file *file){
clear_bit(0,&Device_Open);
MOD_DEC_USE_COUNT;
return 0;
}
it is still racy, that's why struct file_operations and other structs have an 'owner' member......
Signature · 3 lines
Jeff Garzik | Only so many songs can be sung Building 1024 | with two lips, two lungs, and one tongue. MandrakeSoft | - nomeansno
Quoted 4 lines · David C. Hansen
I wrote a quick and dirty char device driver to see if this happened.
If I run two tasks doing a bunch of opens and closes, the-EBUSY
condition in the open function does happen. Is my driver doing something wrong?
What's happening is:
task 1 task 2
-> sys_open
-> lock_kernel
-> testdev_open
-> test_and_set_bit
return success
unlock kernel
-> sys_open
-> lock_kernel (blocks on it)
-> testdev_open
-> test_and_set_bit
return -EBUSY
unlock kernel
returnreturn
The BKL is only held for the duration of the open, not until you close the device.
--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html
Quoted 3 lines · Russell King
The BKL is only held for the duration of the open, not until you close the device.
I'll need to go back and review the changes to the char and block devices. I believe that a big chunk of the drivers that we changed were misc drivers, so I might still be in luck.
Does everyone agree that we need to get the BKL out of common areas like this? For starters, what about adding a pair of spinlocks for block devices and character devices to take the place of the BKL in serializing opens? Or, should we make it the driver's responsibility completely?
Dave Hansen
[email protected]
Quoted 2 lines
Does everyone agree that we need to get the BKL out of common areas like this? For starters, what about adding a pair of spinlocks for block
Ideally the BKL itself should die.
Quoted 3 lines
devices and character devices to take the place of the BKL in serializing opens? Or, should we make it the driver's responsibility completely?
It needs to be the drivers job, to be documented as such and to be implemented properly in some drivers. In paticular there are some extremely interesting closedown races in existing drivers where it goes
CPU1 CPU2
release
[do slow thing]
open
ioctl
setting stuff up
slow thing done
trash the chip setup
turn the chip off
return
ouch bang splat Oops!!!
Alan
Quoted 10 lines · David C. Hansen
Quoted 3 lines · Russell King
The BKL is only held for the duration of the open, not until you close the device.
I'll need to go back and review the changes to the char and block devices. I believe that a big chunk of the drivers that we changed were misc drivers, so I might still be in luck.
Quoted 5 lines
Does everyone agree that we need to get the BKL out of common areas like this? For starters, what about adding a pair of spinlocks for block devices and character devices to take the place of the BKL in serializing opens? Or, should we make it the driver's responsibility completely?
Ideally we should eliminate the BKL completely... but you need to review tons of code when messing around with it, which makes removal annoying. Al Viro seems to know this stuff pretty well, and possibly has plans to remove BKL from various bits of the fs's.
Alan Cox made the comment on IRC that BKL was put into the release method in 2.4 to synchronize open-close-power management. Don't forget to take that third into account either, in the cases where such applies.
Jeff
Signature · 3 lines
Jeff Garzik | Only so many songs can be sung Building 1024 | with two lips, two lungs, and one tongue. MandrakeSoft | - nomeansno
Quoted 4 lines
Alan Cox made the comment on IRC that BKL was put into the release method in 2.4 to synchronize open-close-power management. Don't forget to take that third into account either, in the cases where such applies.
And device unplugging please.
Regards
Oliver
Quoted 5 lines · David C. Hansen
Does everyone agree that we need to get the BKL out of common areas like this? For starters, what about adding a pair of spinlocks for block devices and character devices to take the place of the BKL in serializing opens? Or, should we make it the driver's responsibility completely?
If you do happen to look at the serial driver, please note that I'm currently working on a replacement driver, in cvs at:
:pserver:[email protected]:/mnt/src/cvsroot, module serial (no password)
--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html
Even register_blkdev(), etc hold BKL, without these there will be a lot of problems, all these need to be taken care of if BKL is ever replaced.
Just adding what I know,
Balbir
Quoted 12 lines · Russell King
Quoted 2 lines · David C. Hansen
Nothing, because the BKL is not held for all opens anymore. In most of the cases that we addressed, the BKL was in release _only_, not in open at all. There were quite a few drivers where we added a spinlock, or used atomic operations to keep open from racing with release.
All char and block devs are opened with the BKL held - see
chrdev_openinfs/devices.canddo_openinfs/block_dev.c--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html
Quoted 6 lines
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
-------------------------------------------------------------------------------------------------------------------------
Information transmitted by this E-MAIL is proprietary to Wipro and/or its Customers and is intended for use only by the individual or entity to which it is addressed, and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient or it appears that this mail has been forwarded to you without proper authority, you are notified that any use or dissemination of this information in any manner is strictly prohibited. In such cases, please notify us immediately at mailto:[email protected] and delete this mail from your records.
----------------------------------------------------------------------------------------------------------------------
Even register_blkdev(), etc hold BKL, without these there will be a lot of problems, all these need to be taken care of if BKL is ever replaced.
I think it's pretty clear that the BKL has become so, um, multi-functional over the years that any replacement will not involve a single new variable (or algorithm). On the plus side, it also means it can be a stepwise project. If we carefully identify each function and replace it or obsolete it, we can gradually deprecate the BKL at a safe and sane pace.
Everybody keep track of these usages; there will come a day where somebody is asking you about it again :) open/release is the tip of the iceberg, but it takes a long time for an iceberg to melt ....
Rick
Quoted 2 lines · David C. Hansen
+static
spinlock_tuser_list_lock;
= SPIN_LOCK_UNLOCKED;
Quoted 5 lines · Andrew Morton
Quoted 3 lines · David C. Hansen
+static
spinlock_tuser_list_lock;
=SPIN_LOCK_UNLOCKED;
Whoops. Here's another patch to throw on top of the first one.
arch/i386/kernelapm.c Wed Nov 28 15:37:41 2001 renamed from arch/i386/kernel/apm.c Wed Nov 28 15:37:59 2001 +0 -0
@@ -387,7 +387,7 @@
static DECLARE_WAIT_QUEUE_HEAD(apm_waitqueue); static DECLARE_WAIT_QUEUE_HEAD(apm_suspend_waitqueue);
static struct apm_user * user_list;
+static spinlock_t user_list_lock = SPIN_LOCK_UNLOCKED;
-static spinlock_t user_list_lock;
static char driver_version[] = "1.15"; /* no spaces */
Quoted 1 line · David C. Hansen
Whoops. Here's another patch to throw on top of the first one.
You guys still looking to make global spinlocks static (i.e. not moving to finer-grained locking but just making sure local locks are properly static) ?
My tree has some work to this effect, perhaps I can toss you a patch ...
Robert Love