CC'ing linux-wireless Am Mittwoch, 2. März 2011 schrieb Johannes Berg: > On Wed, 2011-03-02 at 09:53 +0100, Helmut Schaa wrote: > > Am Mittwoch, 2. März 2011 schrieb Johannes Berg: > > > On Wed, 2011-03-02 at 09:37 +0100, Helmut Schaa wrote: > > > > Currently hostapd will force HT Mixed Mode if at least one non-GF STA is > > > > associated. This will force _all_ HT transmissions to be protected. > > > > > > > > 802.11n-2009 doesn't require HT Mixed Mode to be used in case of non-GF > > > > STAs but instead the HT information element contains a flag if non-GF > > > > STAs are present. All STAs are required to protect GF transmissions in > > > > that case. Hence, setting HT Mixed mode if non-GF STAs are present is > > > > superfluous. > > > > > > It seems like this would still be relevant for the AP's TX? > > > > Why? hostapd will configure the new ht_opmode (including the non-GF-STA flag) > > to the driver. And the driver should do the right thing and enable protection only > > for GF transmissions in that case, no? Ok, doing a quick grep through drivers/net/wireless/*: wl1271 doesn't honor this flag right now but seems trivial to fix: 1347 int wl1271_acx_set_ht_information(struct wl1271 *wl, 1348 u16 ht_operation_mode) 1349 { ... 1361 acx->ht_protection = 1362 (u8)(ht_operation_mode & IEEE80211_HT_OP_MODE_PROTECTION); 1363 acx->rifs_mode = 0; 1364 acx->gf_protection = 0; 1365 acx->ht_tx_burst_limit = 0; 1366 acx->dual_cts_protection = 0; rt2800 does the right thing: 1240 static void rt2800_config_ht_opmode(struct rt2x00_dev *rt2x00dev, 1241 struct rt2x00lib_erp *erp) 1242 { 1243 bool any_sta_nongf = !!(erp->ht_opmode & 1244 IEEE80211_HT_OP_MODE_NON_GF_STA_PRSNT); ... 1310 /* check for STAs not supporting greenfield mode */ 1311 if (any_sta_nongf) 1312 gf20_mode = gf40_mode = 2; mwl8k also doesn't honor it, no idea how to fix it: 2603 static int 2604 mwl8k_cmd_set_aid(struct ieee80211_hw *hw, 2605 struct ieee80211_vif *vif, u32 legacy_rate_mask) 2606 { ... 2623 switch (vif->bss_conf.ht_operation_mode & 2624 IEEE80211_HT_OP_MODE_PROTECTION) { 2625 case IEEE80211_HT_OP_MODE_PROTECTION_20MHZ: 2626 prot_mode = MWL8K_FRAME_PROT_11N_HT_40MHZ_ONLY; 2627 break; 2628 case IEEE80211_HT_OP_MODE_PROTECTION_NONHT_MIXED: 2629 prot_mode = MWL8K_FRAME_PROT_11N_HT_ALL; 2630 break; 2631 default: 2632 prot_mode = MWL8K_FRAME_PROT_DISABLED; 2633 break; 2634 } 2635 } 2636 cmd->protection_mode = cpu_to_le16(prot_mode); iwlwifi/iwllegacy seems correct (as long as the fw is doing the right thing): 560 void iwlagn_bss_info_changed(struct ieee80211_hw *hw, 561 struct ieee80211_vif *vif, 562 struct ieee80211_bss_conf *bss_conf, 563 u32 changes) 564 { ... 608 ctx->ht.protection = bss_conf->ht_operation_mode & 609 IEEE80211_HT_OP_MODE_PROTECTION; 610 ctx->ht.non_gf_sta_present = !!(bss_conf->ht_operation_mode & 611 IEEE80211_HT_OP_MODE_NON_GF_STA_PRSNT); 612 iwlagn_check_needed_chains(priv, ctx, bss_conf); 613 iwl_set_rxon_ht(priv, &priv->current_ht_config); Helmut -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html