| Message ID | 20230509015032.3768622-6-tj@kernel.org |
|---|---|
| State | New |
| Headers | show |
| Series | None | expand |
Tejun Heo <tj@kernel.org> wrote: > BACKGROUND > ========== > > When multiple work items are queued to a workqueue, their execution order > doesn't match the queueing order. They may get executed in any order and > simultaneously. When fully serialized execution - one by one in the queueing > order - is needed, an ordered workqueue should be used which can be created > with alloc_ordered_workqueue(). > > However, alloc_ordered_workqueue() was a later addition. Before it, an > ordered workqueue could be obtained by creating an UNBOUND workqueue with > @max_active==1. This originally was an implementation side-effect which was > broken by 4c16bd327c74 ("workqueue: restore WQ_UNBOUND/max_active==1 to be > ordered"). Because there were users that depended on the ordered execution, > 5c0338c68706 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered") > made workqueue allocation path to implicitly promote UNBOUND workqueues w/ > @max_active==1 to ordered workqueues. > > While this has worked okay, overloading the UNBOUND allocation interface > this way creates other issues. It's difficult to tell whether a given > workqueue actually needs to be ordered and users that legitimately want a > min concurrency level wq unexpectedly gets an ordered one instead. With > planned UNBOUND workqueue updates to improve execution locality and more > prevalence of chiplet designs which can benefit from such improvements, this > isn't a state we wanna be in forever. > > This patch series audits all callsites that create an UNBOUND workqueue w/ > @max_active==1 and converts them to alloc_ordered_workqueue() as necessary. > > WHAT TO LOOK FOR > ================ > > The conversions are from > > alloc_workqueue(WQ_UNBOUND | flags, 1, args..) > > to > > alloc_ordered_workqueue(flags, args...) > > which don't cause any functional changes. If you know that fully ordered > execution is not ncessary, please let me know. I'll drop the conversion and > instead add a comment noting the fact to reduce confusion while conversion > is in progress. > > If you aren't fully sure, it's completely fine to let the conversion > through. The behavior will stay exactly the same and we can always > reconsider later. > > As there are follow-up workqueue core changes, I'd really appreciate if the > patch can be routed through the workqueue tree w/ your acks. Thanks. > > Signed-off-by: Tejun Heo <tj@kernel.org> > Cc: Kalle Valo <kvalo@kernel.org> > Cc: "David S. Miller" <davem@davemloft.net> > Cc: Eric Dumazet <edumazet@google.com> > Cc: Jakub Kicinski <kuba@kernel.org> > Cc: Paolo Abeni <pabeni@redhat.com> > Cc: linux-wireless@vger.kernel.org > Cc: netdev@vger.kernel.org I run quick smoke tests with ath11k and ath12k, didn't see any issues. Feel free to take via the workqueue tree: Acked-by: Kalle Valo <kvalo@kernel.org>
diff --git a/drivers/net/wireless/ath/ath10k/qmi.c b/drivers/net/wireless/ath/ath10k/qmi.c index 038c5903c0dc..52c1a3de8da6 100644 --- a/drivers/net/wireless/ath/ath10k/qmi.c +++ b/drivers/net/wireless/ath/ath10k/qmi.c @@ -1082,8 +1082,7 @@ int ath10k_qmi_init(struct ath10k *ar, u32 msa_size) if (ret) goto err; - qmi->event_wq = alloc_workqueue("ath10k_qmi_driver_event", - WQ_UNBOUND, 1); + qmi->event_wq = alloc_ordered_workqueue("ath10k_qmi_driver_event", 0); if (!qmi->event_wq) { ath10k_err(ar, "failed to allocate workqueue\n"); ret = -EFAULT; diff --git a/drivers/net/wireless/ath/ath11k/qmi.c b/drivers/net/wireless/ath/ath11k/qmi.c index ab923e24b0a9..26b252e62909 100644 --- a/drivers/net/wireless/ath/ath11k/qmi.c +++ b/drivers/net/wireless/ath/ath11k/qmi.c @@ -3256,8 +3256,7 @@ int ath11k_qmi_init_service(struct ath11k_base *ab) return ret; } - ab->qmi.event_wq = alloc_workqueue("ath11k_qmi_driver_event", - WQ_UNBOUND, 1); + ab->qmi.event_wq = alloc_ordered_workqueue("ath11k_qmi_driver_event", 0); if (!ab->qmi.event_wq) { ath11k_err(ab, "failed to allocate workqueue\n"); return -EFAULT; diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c index 03ba245fbee9..0a7892b1a8f8 100644 --- a/drivers/net/wireless/ath/ath12k/qmi.c +++ b/drivers/net/wireless/ath/ath12k/qmi.c @@ -3056,8 +3056,7 @@ int ath12k_qmi_init_service(struct ath12k_base *ab) return ret; } - ab->qmi.event_wq = alloc_workqueue("ath12k_qmi_driver_event", - WQ_UNBOUND, 1); + ab->qmi.event_wq = alloc_ordered_workqueue("ath12k_qmi_driver_event", 0); if (!ab->qmi.event_wq) { ath12k_err(ab, "failed to allocate workqueue\n"); return -EFAULT;
BACKGROUND ========== When multiple work items are queued to a workqueue, their execution order doesn't match the queueing order. They may get executed in any order and simultaneously. When fully serialized execution - one by one in the queueing order - is needed, an ordered workqueue should be used which can be created with alloc_ordered_workqueue(). However, alloc_ordered_workqueue() was a later addition. Before it, an ordered workqueue could be obtained by creating an UNBOUND workqueue with @max_active==1. This originally was an implementation side-effect which was broken by 4c16bd327c74 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered"). Because there were users that depended on the ordered execution, 5c0338c68706 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered") made workqueue allocation path to implicitly promote UNBOUND workqueues w/ @max_active==1 to ordered workqueues. While this has worked okay, overloading the UNBOUND allocation interface this way creates other issues. It's difficult to tell whether a given workqueue actually needs to be ordered and users that legitimately want a min concurrency level wq unexpectedly gets an ordered one instead. With planned UNBOUND workqueue updates to improve execution locality and more prevalence of chiplet designs which can benefit from such improvements, this isn't a state we wanna be in forever. This patch series audits all callsites that create an UNBOUND workqueue w/ @max_active==1 and converts them to alloc_ordered_workqueue() as necessary. WHAT TO LOOK FOR ================ The conversions are from alloc_workqueue(WQ_UNBOUND | flags, 1, args..) to alloc_ordered_workqueue(flags, args...) which don't cause any functional changes. If you know that fully ordered execution is not ncessary, please let me know. I'll drop the conversion and instead add a comment noting the fact to reduce confusion while conversion is in progress. If you aren't fully sure, it's completely fine to let the conversion through. The behavior will stay exactly the same and we can always reconsider later. As there are follow-up workqueue core changes, I'd really appreciate if the patch can be routed through the workqueue tree w/ your acks. Thanks. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Kalle Valo <kvalo@kernel.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Eric Dumazet <edumazet@google.com> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Paolo Abeni <pabeni@redhat.com> Cc: linux-wireless@vger.kernel.org Cc: netdev@vger.kernel.org --- drivers/net/wireless/ath/ath10k/qmi.c | 3 +-- drivers/net/wireless/ath/ath11k/qmi.c | 3 +-- drivers/net/wireless/ath/ath12k/qmi.c | 3 +-- 3 files changed, 3 insertions(+), 6 deletions(-)