Re: [Intel-wired-lan] [PATCH V2 net] ice: Re-organizes reqstd/avail {R, T}XQ check/code for efficiency+readability
Dear Salil,
Thank you very much for your patch.
In the git commit message summary, could you please use imperative mood [1]?
Re-organize reqstd/avail {R, T}XQ check/code for efficiency+readability
It’s a bit long though. Maybe:
Avoid unnecessary assignment with user specified {R,T}XQs
Am 14.04.21 um 00:44 schrieb Salil Mehta:
If user has explicitly requested the number of {R,T}XQs, then it is unnecessary to get the count of already available {R,T}XQs from the PF avail_{r,t}xqs bitmap. This value will get overridden by user specified value in any case.
This patch does minor re-organization of the code for improving the flow and readabiltiy. This scope of improvement was found during the review of
readabil*it*y
the ICE driver code.
FYI, I could not test this change due to unavailability of the hardware. It would be helpful if somebody can test this patch and provide Tested-by Tag. Many thanks!
This should go outside the commit message (below the --- for example).
Fixes: 87324e747fde ("ice: Implement ethtool ops for channels")
Did you check the behavior before is actually a bug? Or is it just for the detection heuristic for commits to be applied to the stable series?
Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher jeffrey.t.kirsher@intel.com Signed-off-by: Salil Mehta salil.mehta@huawei.com -- Change V1->V2 (*) Fixed the comments from Anthony Nguyen(Intel) Link: https://lkml.org/lkml/2021/4/12/1997
drivers/net/ethernet/intel/ice/ice_lib.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c index d13c7fc8fb0a..d77133d6baa7 100644 --- a/drivers/net/ethernet/intel/ice/ice_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_lib.c @@ -161,12 +161,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id)
switch (vsi->type) { case ICE_VSI_PF:
vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus());
} else {vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus());
I am curious, did you check the compiler actually creates different code, or did it notice the inefficiency by itself and optimized it already?
pf->num_lan_tx = vsi->alloc_txq;@@ -175,12 +176,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id) if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) { vsi->alloc_rxq = 1; } else {
vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); if (vsi->req_rxq) { vsi->alloc_rxq = vsi->req_rxq; vsi->num_rxq = vsi->req_rxq;
} else {vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); }
Kind regards,
Paul
From: Paul Menzel [mailto:pmenzel@molgen.mpg.de] Sent: Wednesday, April 21, 2021 6:36 AM To: Salil Mehta salil.mehta@huawei.com Cc: linuxarm@openeuler.org; netdev@vger.kernel.org; Linuxarm linuxarm@huawei.com; linux-kernel@vger.kernel.org; Jeff Kirsher jeffrey.t.kirsher@intel.com; intel-wired-lan@lists.osuosl.org; David S. Miller davem@davemloft.net; Jakub Kicinski kuba@kernel.org Subject: Re: [Intel-wired-lan] [PATCH V2 net] ice: Re-organizes reqstd/avail {R, T}XQ check/code for efficiency+readability
Dear Salil,
Thank you very much for your patch.
Thanks for the review.
In the git commit message summary, could you please use imperative mood [1]?
No issues. There is always a scope of improvement.
Re-organize reqstd/avail {R, T}XQ check/code for efficiency+readability
It’s a bit long though. Maybe:
Avoid unnecessary assignment with user specified {R,T}XQs
Umm..above conveys the wrong meaning as this is not what patch is doing.
If you see the code, in the presence of the user specified {R,T}XQs it avoids fetching available {R,T}XQ count.
What about below?
"Avoid unnecessary avail_{r,t}xq assignments if user has specified Qs"
Am 14.04.21 um 00:44 schrieb Salil Mehta:
If user has explicitly requested the number of {R,T}XQs, then it is unnecessary to get the count of already available {R,T}XQs from the PF avail_{r,t}xqs bitmap. This value will get overridden by user specified value in any case.
This patch does minor re-organization of the code for improving the flow and readabiltiy. This scope of improvement was found during the review of
readabil*it*y
Thanks. Missed that earlier. My shaky fingers :(
the ICE driver code.
FYI, I could not test this change due to unavailability of the hardware. It would be helpful if somebody can test this patch and provide Tested-by Tag. Many thanks!
This should go outside the commit message (below the --- for example).
Agreed.
Fixes: 87324e747fde ("ice: Implement ethtool ops for channels")
Did you check the behavior before is actually a bug? Or is it just for the detection heuristic for commits to be applied to the stable series?
Right, later was the idea.
Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher jeffrey.t.kirsher@intel.com Signed-off-by: Salil Mehta salil.mehta@huawei.com -- Change V1->V2 (*) Fixed the comments from Anthony Nguyen(Intel) Link: https://lkml.org/lkml/2021/4/12/1997
drivers/net/ethernet/intel/ice/ice_lib.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c
b/drivers/net/ethernet/intel/ice/ice_lib.c
index d13c7fc8fb0a..d77133d6baa7 100644 --- a/drivers/net/ethernet/intel/ice/ice_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_lib.c @@ -161,12 +161,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16
vf_id)
switch (vsi->type) { case ICE_VSI_PF:
vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus());
} else {vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus());I am curious, did you check the compiler actually creates different code, or did it notice the inefficiency by itself and optimized it already?
I have not looked into that detail but irrespective of what compiler generates I would like to keep the code in a shape which is more efficient and more readable.
I do understand in certain cases we have to do tradeoff between efficiency and readability but I do not see that here.
pf->num_lan_tx = vsi->alloc_txq;@@ -175,12 +176,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16
vf_id)
if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) { vsi->alloc_rxq = 1; } else {
vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); if (vsi->req_rxq) { vsi->alloc_rxq = vsi->req_rxq; vsi->num_rxq = vsi->req_rxq;
} else {vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); }Kind regards,
Paul
[CC: Remove Jeff, as email is rejected]
Dear Salil,
Am 21.04.21 um 09:41 schrieb Salil Mehta:
From: Paul Menzel [mailto:pmenzel@molgen.mpg.de] Sent: Wednesday, April 21, 2021 6:36 AM
[…]
In the git commit message summary, could you please use imperative mood [1]?
No issues. There is always a scope of improvement.
Re-organize reqstd/avail {R, T}XQ check/code for efficiency+readability
It’s a bit long though. Maybe:
Avoid unnecessary assignment with user specified {R,T}XQs
Umm..above conveys the wrong meaning as this is not what patch is doing.
If you see the code, in the presence of the user specified {R,T}XQs it avoids fetching available {R,T}XQ count.
What about below?
"Avoid unnecessary avail_{r,t}xq assignments if user has specified Qs"
Sounds good, still a little long. Maybe:
Avoid unnecessary avail_{r,t}xq assignments with user specified Qs
Am 14.04.21 um 00:44 schrieb Salil Mehta:
If user has explicitly requested the number of {R,T}XQs, then it is unnecessary to get the count of already available {R,T}XQs from the PF avail_{r,t}xqs bitmap. This value will get overridden by user specified value in any case.
This patch does minor re-organization of the code for improving the flow and readabiltiy. This scope of improvement was found during the review of
readabil*it*y
Thanks. Missed that earlier. My shaky fingers :(
the ICE driver code.
FYI, I could not test this change due to unavailability of the hardware. It would be helpful if somebody can test this patch and provide Tested-by Tag. Many thanks!
This should go outside the commit message (below the --- for example).
Agreed.
Fixes: 87324e747fde ("ice: Implement ethtool ops for channels")
Did you check the behavior before is actually a bug? Or is it just for the detection heuristic for commits to be applied to the stable series?
Right, later was the idea.
Cc: intel-wired-lan@lists.osuosl.org Cc: Jeff Kirsher jeffrey.t.kirsher@intel.com Signed-off-by: Salil Mehta salil.mehta@huawei.com -- Change V1->V2 (*) Fixed the comments from Anthony Nguyen(Intel) Link: https://lkml.org/lkml/2021/4/12/1997
drivers/net/ethernet/intel/ice/ice_lib.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c index d13c7fc8fb0a..d77133d6baa7 100644 --- a/drivers/net/ethernet/intel/ice/ice_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_lib.c @@ -161,12 +161,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id)
switch (vsi->type) { case ICE_VSI_PF:
vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus()); if (vsi->req_txq) { vsi->alloc_txq = vsi->req_txq; vsi->num_txq = vsi->req_txq;
} else {vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus()); }I am curious, did you check the compiler actually creates different code, or did it notice the inefficiency by itself and optimized it already?
I have not looked into that detail but irrespective of what compiler generates I would like to keep the code in a shape which is more efficient and more readable.
I do understand in certain cases we have to do tradeoff between efficiency and readability but I do not see that here.
I agree, as *efficiency* is mentioned several times, I assume it was tested. Thank you for the clarification.
pf->num_lan_tx = vsi->alloc_txq;@@ -175,12 +176,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id) if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) { vsi->alloc_rxq = 1; } else {
vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); if (vsi->req_rxq) { vsi->alloc_rxq = vsi->req_rxq; vsi->num_rxq = vsi->req_rxq;
} else {vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); } }
Kind regards,
Paul
From: Paul Menzel [mailto:pmenzel@molgen.mpg.de] Sent: Wednesday, April 21, 2021 8:54 AM
[CC: Remove Jeff, as email is rejected]
Yes, thanks for the reminder. I had noticed it earlier.
[...]
diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c
b/drivers/net/ethernet/intel/ice/ice_lib.c
index d13c7fc8fb0a..d77133d6baa7 100644 --- a/drivers/net/ethernet/intel/ice/ice_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_lib.c @@ -161,12 +161,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi,
u16 vf_id)
switch (vsi->type) { case ICE_VSI_PF:
vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus()); if (vsi->req_txq) { vsi->alloc_txq = vsi->req_txq; vsi->num_txq = vsi->req_txq;
} else {vsi->alloc_txq = min3(pf->num_lan_msix,ice_get_avail_txq_count(pf),(u16)num_online_cpus()); }I am curious, did you check the compiler actually creates different code, or did it notice the inefficiency by itself and optimized it already?
I have not looked into that detail but irrespective of what compiler generates I would like to keep the code in a shape which is more efficient and more readable.
I do understand in certain cases we have to do tradeoff between efficiency and readability but I do not see that here.
I agree, as *efficiency* is mentioned several times, I assume it was tested. Thank you for the clarification.
I mentioned inefficient because below code gets executed unnecessarily.
/** * ice_get_avail_q_count - Get count of queues in use * @pf_qmap: bitmap to get queue use count from * @lock: pointer to a mutex that protects access to pf_qmap * @size: size of the bitmap */ static u16 ice_get_avail_q_count(unsigned long *pf_qmap, struct mutex *lock, u16 size) { unsigned long bit; u16 count = 0;
mutex_lock(lock); for_each_clear_bit(bit, pf_qmap, size) count++; mutex_unlock(lock);
return count; }
/** * ice_get_avail_txq_count - Get count of Tx queues in use * @pf: pointer to an ice_pf instance */ u16 ice_get_avail_txq_count(struct ice_pf *pf) { return ice_get_avail_q_count(pf->avail_txqs, &pf->avail_q_mutex, pf->max_pf_txqs); }
pf->num_lan_tx = vsi->alloc_txq;@@ -175,12 +176,13 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi,
u16 vf_id)
if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) { vsi->alloc_rxq = 1; } else {
vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); if (vsi->req_rxq) { vsi->alloc_rxq = vsi->req_rxq; vsi->num_rxq = vsi->req_rxq;
} else {vsi->alloc_rxq = min3(pf->num_lan_msix,ice_get_avail_rxq_count(pf),(u16)num_online_cpus()); } }Kind regards,
Paul
-
Paul Menzel -
Salil Mehta