dim_calc_stats() computes the per-millisecond rates as
DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us)
where nbytes is a u32 and USEC_PER_MSEC is 1000L. On 64-bit the product
is done in 64-bit long arithmetic, but on 32-bit architectures long is
32 bits wide and the product wraps as soon as a measurement window
carries more than 4294967 bytes (about 4.3 MB). The same applies to the
packet and completion counts, although those need more than 4.29
million packets or completions per window.
A DIM window spans DIM_NEVENTS (64) events. Drivers count events per
interrupt or per NAPI poll, so under sustained load a window can easily
carry more than 4.3 MB: 64 full NAPI polls of 64 MTU-sized frames are
already 6.2 MB, and drivers such as mtk_eth_soc count one event per
interrupt while NAPI keeps polling with the interrupt masked. On 32-bit
users of the library (for example mtk_eth_soc on MT7621, bcmgenet and
bcmsysport on 32-bit ARM, or virtio_net in a 32-bit guest) bpms then
becomes the product modulo 2^32 divided by the window length, and
net_dim_stats_compare() makes its BETTER/WORSE decisions on a value
that has little to do with the real throughput.
For example, a 1 Gbit/s link at line rate that moves 5 MB in a 40 ms
window gives bpms = 125000 on 64-bit but 17626 on 32-bit, and 5 million
packets in 2 s gives ppms = 353 instead of 2500.
Widen the products to 64 bits and divide with DIV_ROUND_UP_ULL(). The
results are unchanged on 64-bit.
Fixes: cb3c7fd4f8 ("net/mlx5e: Support adaptive RX coalescing")
Fixes: 4c4dbb4a73 ("net/mlx5e: Move dynamic interrupt coalescing code to include/linux")
Assisted-by: LLM
Signed-off-by: Shashank Mohan Jain <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Allow the Dynamic Interrupt Moderation (DIM) library to be built as a
module. This is particularly useful in an Android GKI (Google Kernel
Image) configuration where everything is built as a module, including
Ethernet controller drivers. Having to build DIMLIB into the kernel
image with potentially no user is wasteful.
Signed-off-by: Florian Fainelli <[email protected]>
Reviewed-by: Alexander Lobakin <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Add return value for dim_calc_stats. This is an indication for the
caller if curr_stats was assigned by the function. Avoid using
curr_stats uninitialized over {rdma/net}_dim, when no time delta between
samples. Coverity reported this potential use of an uninitialized
variable.
Fixes: 4c4dbb4a73 ("net/mlx5e: Move dynamic interrupt coalescing code to include/linux")
Fixes: cb3c7fd4f8 ("net/mlx5e: Support adaptive RX coalescing")
Signed-off-by: Roy Novich <[email protected]>
Reviewed-by: Aya Levin <[email protected]>
Reviewed-by: Saeed Mahameed <[email protected]>
Signed-off-by: Tariq Toukan <[email protected]>
Reviewed-by: Leon Romanovsky <[email protected]>
Reviewed-by: Michal Kubiak <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Paolo Abeni <[email protected]>
While using net_dim, a dim_sample was used without ever initializing the
comps value. Added use of DIV_ROUND_DOWN_ULL() to prevent potential
overflow, it should not be a problem to save the final result in an int
because after the division by epms the value should not be larger than a
few thousand.
[ 1040.127124] UBSAN: Undefined behaviour in lib/dim/dim.c:78:23
[ 1040.130118] signed integer overflow:
[ 1040.131643] 134718714 * 100 cannot be represented in type 'int'
Fixes: 398c2b05bb ("linux/dim: Add completions count to dim_sample")
Signed-off-by: Yamin Friedman <[email protected]>
Signed-off-by: Leon Romanovsky <[email protected]>
Acked-by: Saeed Mahameed <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Added a measurement of completions per/msec to allow for completion based
dim algorithms.
In order to use dynamic interrupt moderation with RDMA we need to have a
different measurment than packets per second. This change is meant to
prepare for adding a new DIM method.
All drivers that use net_dim and thus do not need a completion count will
have the completions set to 0.
Signed-off-by: Yamin Friedman <[email protected]>
Reviewed-by: Max Gurtovoy <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
Moved all logic from dim.h and net_dim.h to dim.c and net_dim.c.
This is both more structurally appealing and would allow to only
expose externally used functions.
Signed-off-by: Tal Gilboa <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>
Shashank Mohan Jain