On 6/15/2023 4:51 AM, Kalle Valo wrote:
Aloka Dixit <quic_alokad@xxxxxxxxxxx> writes:
Propagate EHT capabilities to the userspace using a new member
'eht_cap' in structure ieee80211_sband_iftype_data.
MCS-NSS capabilities are copied depending on the supported bandwidths
for the given band.
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Aloka Dixit <quic_alokad@xxxxxxxxxxx>
Signed-off-by: Pradeep Kumar Chitrapu<quic_pradeepc@xxxxxxxxxxx>
[...]
+static void ath12k_mac_copy_eht_ppet_ru(u32 ppet, u8 ppe_thres[], int ru)
+{
+ int i;
+ u32 val = 0;
+ u8 ppet_size_ru = IEEE80211_EHT_PPE_THRES_INFO_PPET_SIZE * 2;
+ u8 bit = IEEE80211_EHT_PPE_THRES_INFO_HEADER_SIZE;
+
+ u32p_replace_bits(&val, ppet >> (ru * ppet_size_ru),
+ GENMASK(ppet_size_ru - 1, 0));
+
+ val = ((val >> IEEE80211_EHT_PPE_THRES_INFO_PPET_SIZE) & GENMASK(2, 0)) |
+ ((val & GENMASK(2, 0)) << IEEE80211_EHT_PPE_THRES_INFO_PPET_SIZE);
This shifting of val looks weird. I didn't check the spec, what does it
do? Is there any cleaner way to do this? And should have a define for
GENMASK(2, 0)?
I have a follow-up ready, based on top of the fixes you already made in
the master pending branch, which cleans this up.
Also noticed that need to use hweight16 instead of hweight8 for
IEEE80211_EHT_PPE_THRES_RU_INDEX_BITMASK_MASK. The new version fixes it
in this patch as well as in patch #6.
Should I send the next version?
