Hello, While studying the Linux source code (2.4.8) for x86, I noticed that sleep_on() and wait_event() behave differently with respect to the eflags register: - sleep_on() saves the flags [by using wq_write_lock_irqsave() in the SLEEP_ON_HEAD macro] before calling schedule() and restores them [by using wq_write_unlock_irqrestore() in the SLEEP_ON_TAIL macro] when returning from schedule. - wait_event() does not save/restore the eflags register before/after schedule. Since schedule() does not save/restore the eflags register of the previous/next processes (at least I couldn't find any corresponding code in the switch_to() and __switch_to() macros). This results in a different behaviour, in that a process may assume the eflags to remain constant across calls to sleep_on(), while they may be modified (by another process) across calls to wait_event(). So my questions are: 1) Are my observations correct. In particular is it true, that the eflags are not saved/restored by schedule()? 2) Why did the kernel developpers choose to implement a different semantic (with respect to the eflags register) for sleep_on() and wait_event()? 3) Are such differences documented at any place, or is it up to the developper of a kernel module, to figure out such differences from the Linux sources? thanks in advance for any help with best regards Martin P.S. please put me on cc: in your reply, since I am not in the mailing list. -- Supercomputing System AG email: maletinsky@scs.ch Martin Maletinsky phone: +41 (0)1 445 16 05 Technoparkstrasse 1 fax: +41 (0)1 445 16 10 CH-8005 Zurich -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
