kernel/drivers/net/ethernet/qualcomm/rmnet, branch linux-5.14.y

net: qualcomm: rmnet: fix two pointer math bugs

2021-06-21T19:19:19Z

We recently changed these two pointers from void pointers to struct pointers and it breaks the pointer math so now the "txphdr" points beyond the end of the buffer. Fixes: 56a967c4f7e5 ("net: qualcomm: rmnet: Remove some unneeded casts") Signed-off-by: Dan Carpenter Reviewed-by: Subash Abhinov Kasiviswanathan Signed-off-by: David S. Miller

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

2021-06-19T02:47:02Z

Trivial conflicts in net/can/isotp.c and tools/testing/selftests/net/mptcp/mptcp_connect.sh scaled_ppm_to_ppb() was moved from drivers/ptp/ptp_clock.c to include/linux/ptp_clock_kernel.h in -next so re-apply the fix there. Signed-off-by: Jakub Kicinski

net: qualcomm: rmnet: Remove some unneeded casts

2021-06-16T19:19:50Z

Remove the explicit casts in the checksum complement functions and pass the actual protocol specific headers instead. Signed-off-by: Subash Abhinov Kasiviswanathan Signed-off-by: David S. Miller

net: qualcomm: rmnet: Allow partial updates of IFLA_FLAGS

2021-06-16T19:16:22Z

The idiomatic way to handle the changelink flags/mask pair seems to be allow partial updates of the driver's link flags. In contrast the rmnet driver masks the incoming flags and then use that as the new flags. Change the rmnet driver to follow the common scheme, before the introduction of IFLA_RMNET_FLAGS handling in iproute2 et al. Signed-off-by: Bjorn Andersson Reviewed-by: Alex Elder Reviewed-by: Subash Abhinov Kasiviswanathan Signed-off-by: David S. Miller

net: qualcomm: rmnet: always expose a few functions

2021-06-14T19:43:45Z

A recent change tidied up some conditional code, avoiding the use of some #ifdefs. Unfortunately, if CONFIG_IPV6 was not enabled, it meant that two functions were referenced but never defined. The easiest fix is to just define stubs for these functions if CONFIG_IPV6 is not defined. This will soon be simplified further by some other development in the works... Reported-by: kernel test robot Fixes: 75db5b07f8c39 ("net: qualcomm: rmnet: eliminate some ifdefs") Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: qualcomm: rmnet: don't over-count statistics

2021-06-14T19:13:38Z

The purpose of the loop using u64_stats_fetch_*_irq() is to ensure statistics on a given CPU are collected atomically. If one of the statistics values gets updated within the begin/retry window, the loop will run again. Currently the statistics totals are updated inside that window. This means that if the loop ever retries, the statistics for the CPU will be counted more than once. Fix this by taking a snapshot of a CPU's statistics inside the protected window, and then updating the counters with the snapshot values after exiting the loop. (Also add a newline at the end of this file...) Fixes: 192c4b5d48f2a ("net: qualcomm: rmnet: Add support for 64 bit stats") Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: qualcomm: rmnet: IPv6 payload length is simple

2021-06-12T20:18:49Z

We don't support any extension headers for IPv6 packets. Extension headers therefore contribute 0 bytes to the payload length. As a result we can just use the IPv6 payload length as the length used to compute the pseudo header checksum for both UDP and TCP messages. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: qualcomm: rmnet: drop some unary NOTs

2021-06-12T20:18:49Z

We compare a payload checksum with a pseudo checksum value for equality in rmnet_map_ipv4_dl_csum_trailer(). Both of those values are computed with a unary NOT (~) operation. The result of the comparison is the same if we omit that NOT for both values. Remove these operations in rmnet_map_ipv6_dl_csum_trailer() also. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: qualcomm: rmnet: trailer value is a checksum

2021-06-12T20:18:49Z

The csum_value field in the rmnet_map_dl_csum_trailer structure is a "real" Internet checksum. It is a 16 bit value, in big endian format, which represents an inverted ones' complement sum over pairs of bytes. Make that clear by changing its type to __sum16. This makes a typecast in rmnet_map_ipv4_dl_csum_trailer() and another in rmnet_map_ipv6_dl_csum_trailer() unnecessary. Signed-off-by: Alex Elder Signed-off-by: David S. Miller

net: qualcomm: rmnet: remove unneeded code

2021-06-12T20:18:48Z

The previous patch makes rmnet_map_ipv4_dl_csum_trailer() return early with an error if it is determined that the computed checksum for the IP payload does not match what was expected. If the computed checksum *does* match the expected value, the IP payload (i.e., the transport message), can be considered good. There is no need to do any further processing of the message. This means a big block of code is unnecessary for validating the transport checksum value, and can be removed. Make comparable changes in rmnet_map_ipv6_dl_csum_trailer(). Signed-off-by: Alex Elder Signed-off-by: David S. Miller