The OpenNET Project / Index page

[ новости /+++ | форум | теги | ]



Индекс форумов
Составление сообщения

Исходное сообщение
"Релиз ядра Linux 3.15"
Отправлено pavlinux, 09-Июн-14 01:45 
Merge futex fixes from Thomas Gleixner:
"So with more awake and less futex wreckaged brain, I went through my
  list of points again and came up with the following 4 patches.

  1) Prevent pi requeueing on the same futex

     I kept Kees check for uaddr1 == uaddr2 as a early check for private
     futexes and added a key comparison to both futex_requeue and
     futex_wait_requeue_pi.

     Sebastian, sorry for the confusion yesterday night.  I really
     misunderstood your question.

     You are right the check is pointless for shared futexes where the
     same physical address is mapped to two different virtual addresses.

  2) Sanity check atomic acquisiton in futex_lock_pi_atomic

     That's basically what Darren suggested.

     I just simplified it to use futex_top_waiter() to find kernel
     internal state.  If state is found return -EINVAL and do not bother
     to fix up the user space variable.  It's corrupted already.

  3) Ensure state consistency in futex_unlock_pi

     The code is silly versus the owner died bit.  There is no point to
     preserve it on unlock when the user space thread owns the futex.

     What's worse is that it does not update the user space value when
     the owner died bit is set.  So the kernel itself creates observable
     inconsistency.

     Another "optimization" is to retry an atomic unlock.  That's
     pointless as in a sane environment user space would not call into
     that code if it could have unlocked it atomically.  So we always
     check whether there is kernel state around and only if there is
     none, we do the unlock by setting the user space value to 0.

  4) Sanitize lookup_pi_state

     lookup_pi_state is ambigous about TID == 0 in the user space value.

     This can be a valid state even if there is kernel state on this
     uaddr, but we miss a few corner case checks.

     I tried to come up with a smaller solution hacking the checks into
     the current cruft, but it turned out to be ugly as hell and I got
     more confused than I was before.  So I rewrote the sanity checks
     along the state documentation with awful lots of commentry"

* emailed patches from Thomas Gleixner <tglx@linutronix.de>:
  futex: Make lookup_pi_state more robust
  futex: Always cleanup owner tid in unlock_pi
  futex: Validate atomic acquisition in futex_lock_pi_atomic()
  futex-prevent-requeue-pi-on-same-futex.patch futex: Forbid uaddr == uaddr2 in futex_requeue(..., requeue_pi=1)

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux....

 

Ваше сообщение
Имя*:
EMail:
Для отправки ответов на email укажите знак ! перед адресом, например, !user@host.ru (!! - не показывать email).
Более тонкая настройка отправки ответов производится в профиле зарегистрированного участника форума.
Заголовок*:
Сообщение*:
 
При общении не допускается: неуважительное отношение к собеседнику, хамство, унизительное обращение, ненормативная лексика, переход на личности, агрессивное поведение, обесценивание собеседника, провоцирование флейма голословными и заведомо ложными заявлениями. Не отвечайте на сообщения, явно нарушающие правила - удаляются не только сами нарушения, но и все ответы на них. Лог модерирования.



Партнёры:
PostgresPro
Inferno Solutions
Hosting by Hoster.ru
Хостинг:

Закладки на сайте
Проследить за страницей
Created 1996-2024 by Maxim Chirkov
Добавить, Поддержать, Вебмастеру