From: Tim Sell Date: Wed, 1 Feb 2017 22:38:58 +0000 (-0500) Subject: staging: unisys: visornic: prevent hang doing 'modprobe -r visornic' X-Git-Tag: v4.11-rc1~116^2~210 X-Git-Url: https://asedeno.scripts.mit.edu/gitweb/?a=commitdiff_plain;h=9c70ee32ffedc48705b08d97212683a5482628cb;p=linux.git staging: unisys: visornic: prevent hang doing 'modprobe -r visornic' A stray+extraneous 'netif_napi_add()' that we were doing in visornic_probe() was causing havoc when we got into visornic_remove(), called during 'modprobe -r visornic'. The symptom was a processor busy-wait loop on the modprobe process, which '/proc//stack' would show looping doing napi things. Presumably the stray line got there as a result of some merging snafoo, and has been deleted to fix the problem. With this patch 'modprobe -r visornic' and a subsequent 'modprobe visornic' both complete successfully, and result in an operational network. Signed-off-by: Tim Sell Signed-off-by: David Kershner Signed-off-by: Greg Kroah-Hartman --- diff --git a/drivers/staging/unisys/visornic/visornic_main.c b/drivers/staging/unisys/visornic/visornic_main.c index c44c430b966f..d8f5ecac6c12 100644 --- a/drivers/staging/unisys/visornic/visornic_main.c +++ b/drivers/staging/unisys/visornic/visornic_main.c @@ -1802,7 +1802,7 @@ static int visornic_probe(struct visor_device *dev) /* TODO: Setup Interrupt information */ /* Let's start our threads to get responses */ - netif_napi_add(netdev, &devdata->napi, visornic_poll, 64); + netif_napi_add(netdev, &devdata->napi, visornic_poll, NAPI_WEIGHT); setup_timer(&devdata->irq_poll_timer, poll_for_irq, (unsigned long)devdata); @@ -1832,9 +1832,6 @@ static int visornic_probe(struct visor_device *dev) goto cleanup_napi_add; } - /* Let's start our threads to get responses */ - netif_napi_add(netdev, &devdata->napi, visornic_poll, NAPI_WEIGHT); - /* Note: Interrupts have to be enable before the while * loop below because the napi routine is responsible for * setting enab_dis_acked