From 24f4649cd8fcbb9ab988933890257f450260fd3d Mon Sep 17 00:00:00 2001 From: Zhang Yuchen Date: Fri, 7 Oct 2022 17:26:16 +0800 Subject: [PATCH] ipmi: fix long wait in unload when IPMI disconnect commit f6f1234d98cce69578bfac79df147a1f6660596c upstream. When fixing the problem mentioned in PATCH1, we also found the following problem: If the IPMI is disconnected and in the sending process, the uninstallation driver will be stuck for a long time. The main problem is that uninstalling the driver waits for curr_msg to be sent or HOSED. After stopping tasklet, the only place to trigger the timeout mechanism is the circular poll in shutdown_smi. The poll function delays 10us and calls smi_event_handler(smi_info,10). Smi_event_handler deducts 10us from kcs->ibf_timeout. But the poll func is followed by schedule_timeout_uninterruptible(1). The time consumed here is not counted in kcs->ibf_timeout. So when 10us is deducted from kcs->ibf_timeout, at least 1 jiffies has actually passed. The waiting time has increased by more than a hundredfold. Now instead of calling poll(). call smi_event_handler() directly and calculate the elapsed time. For verification, you can directly use ebpf to check the kcs-> ibf_timeout for each call to kcs_event() when IPMI is disconnected. Decrement at normal rate before unloading. The decrement rate becomes very slow after unloading. $ bpftrace -e 'kprobe:kcs_event {printf("kcs->ibftimeout : %d\n", *(arg0+584));}' Signed-off-by: Zhang Yuchen Message-Id: <20221007092617.87597-3-zhangyuchen.lcr@bytedance.com> Signed-off-by: Corey Minyard Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman --- drivers/char/ipmi/ipmi_si_intf.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) diff --git a/drivers/char/ipmi/ipmi_si_intf.c b/drivers/char/ipmi/ipmi_si_intf.c index 6f3272b..17255e7 100644 --- a/drivers/char/ipmi/ipmi_si_intf.c +++ b/drivers/char/ipmi/ipmi_si_intf.c @@ -2152,6 +2152,20 @@ skip_fallback_noirq: } module_init(init_ipmi_si); +static void wait_msg_processed(struct smi_info *smi_info) +{ + unsigned long jiffies_now; + long time_diff; + + while (smi_info->curr_msg || (smi_info->si_state != SI_NORMAL)) { + jiffies_now = jiffies; + time_diff = (((long)jiffies_now - (long)smi_info->last_timeout_jiffies) + * SI_USEC_PER_JIFFY); + smi_event_handler(smi_info, time_diff); + schedule_timeout_uninterruptible(1); + } +} + static void shutdown_smi(void *send_info) { struct smi_info *smi_info = send_info; @@ -2186,16 +2200,13 @@ static void shutdown_smi(void *send_info) * in the BMC. Note that timers and CPU interrupts are off, * so no need for locks. */ - while (smi_info->curr_msg || (smi_info->si_state != SI_NORMAL)) { - poll(smi_info); - schedule_timeout_uninterruptible(1); - } + wait_msg_processed(smi_info); + if (smi_info->handlers) disable_si_irq(smi_info); - while (smi_info->curr_msg || (smi_info->si_state != SI_NORMAL)) { - poll(smi_info); - schedule_timeout_uninterruptible(1); - } + + wait_msg_processed(smi_info); + if (smi_info->handlers) smi_info->handlers->cleanup(smi_info->si_sm); -- 2.7.4