March 2015 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

Question on mutex code
by Matthias Bonne March 15, 2015

March 15, 2015

Hello, I am trying to understand how mutexes work in the kernel, and I think there might be a race between mutex_trylock() and mutex_unlock(). More specifically, the race is between the functions __mutex_trylock_slowpath and __mutex_unlock_common_slowpath (both defined in kernel/locking/mutex.c). Consider the following sequence of events: 0. Suppose a mutex is locked by task A and has no waiters. 1. Task B calls mutex_trylock(). 2. mutex_trylock() calls the architecture-specific __mutex_fastpath_trylock(), with __mutex_trylock_slowpath() as fail_fn. 3. According to the description of __mutex_fastpath_trylock() (for example in include/asm-generic/mutex-dec.h), "if the architecture has no effective trylock variant, it should call the fail_fn spinlock-based trylock variant unconditionally". So __mutex_fastpath_trylock() may now call __mutex_trylock_slowpath(). 4. Task A releases the mutex. 5. Task B, in __mutex_trylock_slowpath, executes: /* No need to trylock if the mutex is locked. */ if (mutex_is_locked(lock)) return 0; Since the mutex is no longer locked, the function continues. 6. Task C, which runs on a different cpu than task B, locks the mutex again. 7. Task B, in __mutex_trylock_slowpath(), continues: spin_lock_mutex(&lock->wait_lock, flags); prev = atomic_xchg(&lock->count, -1); if (likely(prev == 1)) { mutex_set_owner(lock); mutex_acquire(&lock->dep_map, 0, 1, _RET_IP_); } At this point task B holds mutex->wait_lock, prev is 0 (because there are no waiters other than task B, so the count was 0) and the mutex count is set to -1. 5. Task C calls mutex_unlock() to unlock the mutex. 6. mutex_unlock() calls the architecture-specific function __mutex_fastpath_unlock(), which fails (because the mutex count is -1), so it now calls __mutex_unlock_slowpath(), which calls __mutex_unlock_common_slowpath(). 7. __mutex_unlock_common_slowpath() sets the mutex count to 1 unconditionally, before spinning on mutex->wait_lock. 8. Task B, in __mutex_trylock_slowpath, continues: /* Set it back to 0 if there are no waiters: */ if (likely(list_empty(&lock->wait_list))) atomic_set(&lock->count, 0); spin_unlock_mutex(&lock->wait_lock, flags); return prev == 1; mutex->wait_list is still empty, so the code sets the mutex count to zero (which means the mutex is locked), releases mutex->wait_lock, and returns 0 (which means that the mutex is locked by someone else, and cannot be acquired). 9. Task C, in __mutex_unlock_common_slowpath, acquires mutex->wait_lock, unlocks it immediately (because there are no waiters to wake up) and returns. The end result is that the mutex count is 0 (locked), although the owner has just released it, and nobody else is holding the mutex. So it can no longer be acquired by anyone. Am I missing something that prevents the above scenario from happening? If not, should I post a patch that fixes it to LKML? Or is it considered too "theoretical" and cannot happen in practice? Thanks.

4 6

CHECK: Alignment should match open parenthesis
by palik imre March 13, 2015

March 13, 2015

Sorry for the silly question, but I have some issues with this checkpatch.pl warning. I mean Documentation/CodingStyle says: Outside of comments, documentation and except in Kconfig, spaces are never used for indentation, and the above example is deliberately broken. But checkpatch.pl claims I should align to open parentheses. These two things seem to be contradictory to me. Could somebody clarify this? Thanks, Imre

5 14

Re:question scheduler
by Meyer Lansky March 13, 2015

March 13, 2015

hello I do not understand that these two lines do pcred = __task_cred(p); match = (cred->euid == pcred->euid || cred->euid == pcread->uid); or what they mean ? excuse me for my bad english .

1 0

Help with learning scheduler O(1)
by Meyer Lansky March 13, 2015

March 13, 2015

Hello all continue to study the scheduler is O(1 ) there was one difficulty with the string 2 on 3, in that these blocking occurs after lines ? a reference to the code that study - http://lxr.free-electrons.com/source/kernel/sched.c?v=2.6.36 what I learn on line 4682 to 4695. 1 rcu_read_lock(); 2 pcred = __task_cred(p); 3 match = (cred->euid == pcred->euid || cred->euid == pcread->uid); 4 rcu_read_unlock(); I ask because in the comments there is not described . Thank.

2 1

Missing sys_enter_open events
by sahil aggarwal March 13, 2015

March 13, 2015

Hi all I am doing ftrace to track files being opened and mapped by process but sys_enter_open is getting missed and getname is being traced which is called by sys_open only. What is happening.? mem-2474 [005] 681774.985957: getnameprobe: (sys_execve+0x21/0x5a <- getname) arg1="bin/mem" mem-2474 [001] 683183.894867: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=80 mem-2474 [001] 683183.894900: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=1c0 mem-2474 [001] 683183.894902: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=1c mem-2474 [001] 683183.894913: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=80 mem-2474 [001] 683183.895113: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=188 mem-2474 [001] 683183.895275: getnameprobe: (do_sys_open+0x3b/0x105 <- getname) arg1="/etc/ld.so.cache" mem-2474 [001] 683183.895316: getnameprobe: (do_sys_open+0x3b/0x105 <- getname) arg1="/lib/libc.so.6" mem-2474 [001] 683183.895321: sys_read(fd: 4, buf: 7fff8461a0f8, count: 340) mem-2474 [001] 683183.895328: sync_read: (vfs_read+0xab/0x107 <- do_sync_read) arg1=340

1 0

I am new here
by Leonardo Ariel Carreras Rodríguez March 12, 2015

March 12, 2015

Hello! I am new here, and I want to know how to get started, maybe with the simplest way available. Regards, Leonardo

6 8

Re: I am new here
by Jonathan Jin March 12, 2015

March 12, 2015

Leonardo Ariel Carreras Rodríguez <zc77leo(a)gmail.com> writes: > In particular, I am asking about how much time do you have to wait > until you get accepted to join the challenge. If s/he has to approve > each mail address, this will take a lot of time. Again, it varies. Challenge joins shouldn't take too long, as I imagine that processing those emails was one of the first jobs that got automated. If I remember correctly, my registration had a response time on the order of a couple of hours. Side note: don't top-post. It's very frowned upon in most mailing lists. See here: http://kernelnewbies.org/mailinglistguidelines. -- Jonathan Jin jjin.me github.com/jinnovation linkedin.com/in/jonathanjin

1 0

RCU is not safe enough to protect __d_lookup ?
by Rock Lee March 12, 2015

March 12, 2015

Hi, all: I am studying the dcache in VFS recently . I've found that hlist of dcache is already protected by RCU in __d_lookup.Is it necessary for the function -- d_lookup using sequence lock to protect the hlist again? struct dentry * d_lookup(struct dentry * parent, struct qstr * name){ struct dentry * dentry = NULL; unsigned long seq; do { seq = read_seqbegin(&rename_lock); dentry = __d_lookup(parent, name); if (dentry) break; } while (read_seqretry(&rename_lock, seq)); return dentry;} Thanks in advance--------------------------------Rock Lee

1 0

RCU is not safe enough to protect __d_lookup ?
by Rock Lee March 12, 2015

March 12, 2015

Hi, all: I am studying the dcache in VFS recently . I've found that hlist of dcache is already protected by RCU in __d_lookup.Is it necessary for the function -- d_lookup using sequence lock to protect the hlist again? struct dentry * d_lookup(struct dentry * parent, struct qstr * name){ struct dentry * dentry = NULL; unsigned long seq; do { seq = read_seqbegin(&rename_lock); dentry = __d_lookup(parent, name); if (dentry) break; } while (read_seqretry(&rename_lock, seq)); return dentry;} Thanks in advance--------------------------------Rock Lee

1 0

Tracing allocators of virtual memory and main memory
by sahil aggarwal March 12, 2015

March 12, 2015

Hi all If i want to trace the allocators of virtual memory and main memory of process using ftrace on which functions i need to enable the probe.? Currently i have enabled sys_mmap,sys_brk for virtual allocations and mm_page_alloc, kmalloc, kmem_cache_alloc for main memory allocation. Will this give me whole picture of virtual and main memory allocations.? Thanks Regards Sahil

4 5