From: James Prestwood <prestwoj@gmail.com>
To: iwd@lists.linux.dev
Subject: Re: [PATCH v3 07/10] station: roam blacklist BSS's, and consider when roaming
Date: Wed, 26 Mar 2025 10:49:57 -0700 [thread overview]
Message-ID: <d311d92a-b236-4ff1-89bc-377a200e4cb4@gmail.com> (raw)
In-Reply-To: <20250325180041.238676-7-prestwoj@gmail.com>
On 3/25/25 11:00 AM, James Prestwood wrote:
> If the BSS is requesting IWD roam elsewhere add this BSS to the
> blacklist using BLACKLIST_REASON_ROAM_REQUESTED. This will lower
> the chances of IWD roaming/connecting back to this BSS in the
> future.
>
> This then allows IWD to consider this blacklist state when picking
> a roam candidate. Its undesireable to fully ban a roam blacklisted
> BSS, so some additional sorting logic has been added. Prior to
> comparing based on rank, BSS's will be sorted into 3 groups:
>
> Optimal - Not roam blacklisted, and above the roaming threshold
> Above Threshold - May be blacklisted, and above the roaming threshold
> Below Threshold - BSS is below the roaming threshold
> ---
> src/station.c | 88 ++++++++++++++++++++++++++++++++++++++++++++++++---
> 1 file changed, 84 insertions(+), 4 deletions(-)
>
> diff --git a/src/station.c b/src/station.c
> index e2ed78f3..e3c7c189 100644
> --- a/src/station.c
> +++ b/src/station.c
> @@ -155,6 +155,57 @@ struct anqp_entry {
> uint32_t pending;
> };
>
> +/*
> + * Rather than sorting BSS's purely based on ranking a higher level grouping
> + * is used. The purpose of this higher order grouping is the consider the BSS's
> + * roam blacklist status. The roam blacklist is a "soft" blacklist in that we
> + * still should connect to these BSS's if they are the only "good" option.
> + * The question here is: what makes a BSS "good" vs "bad".
> + *
> + * For an initial (probably nieve) approach here we can use the RoamThreshod[5G]
> + * and sort BSS's into "above" or "below" this threshold. Within each of these
> + * groups a BSS may be blacklisted, meaning it should get sorted lower on the
> + * list compared to others within the same group.
> + *
> + * Since we have several call sites needing to check/compare these groupings
> + * a bitmask can be used to describe the grouping:
> + *
> + * Each group will have 2 bits. The lower order bit signifies the BSS is in the
> + * group, but also blacklisted. The higher order bit signifies the BSS is in
> + * the group, but not blacklisted. The bitmask between two BSS's can then be
> + * compared directly similar to rank.
> + */
> +
> +#define UNDER_THRESHOLD_BIT 1
> +#define ABOVE_THRESHOLD_BIT 3
> +
> +static uint32_t evaluate_bss_group(const uint8_t *addr, uint32_t freq,
> + int16_t signal_strength)
> +{
> + int threshold;
> + int signal = signal_strength / 100;
> + bool roam_blacklist;
> + uint32_t mask = 0;
> +
> + if (blacklist_contains_bss(addr, BLACKLIST_REASON_CONNECT_FAILED))
> + return mask;
> +
> + roam_blacklist = blacklist_contains_bss(addr,
> + BLACKLIST_REASON_ROAM_REQUESTED);
> +
> + if (freq > 4000)
> + netdev_get_low_signal_thresholds(NULL, &threshold);
> + else
> + netdev_get_low_signal_thresholds(&threshold, NULL);
> +
> + if (signal >= threshold)
> + set_bit(&mask, ABOVE_THRESHOLD_BIT - roam_blacklist);
> + else
> + set_bit(&mask, UNDER_THRESHOLD_BIT - roam_blacklist);
> +
> + return mask;
> +}
> +
> /*
> * Used as entries for the roam list since holding scan_bss pointers directly
> * from station->bss_list is not 100% safe due to the possibility of the
> @@ -164,11 +215,13 @@ struct roam_bss {
> uint8_t addr[6];
> uint16_t rank;
> int32_t signal_strength;
> + uint32_t group;
> bool ft_failed: 1;
> };
>
> static struct roam_bss *roam_bss_from_scan_bss(const struct scan_bss *bss,
> - uint16_t rank)
> + uint16_t rank,
> + uint32_t group)
> {
> struct roam_bss *rbss = l_new(struct roam_bss, 1);
>
> @@ -176,6 +229,7 @@ static struct roam_bss *roam_bss_from_scan_bss(const struct scan_bss *bss,
> rbss->rank = rank;
> rbss->signal_strength = bss->signal_strength;
> rbss->ft_failed = false;
> + rbss->group = group;
>
> return rbss;
> }
> @@ -184,6 +238,12 @@ static int roam_bss_rank_compare(const void *a, const void *b, void *user_data)
> {
> const struct roam_bss *new_bss = a, *bss = b;
>
> + if (bss->group > new_bss->group)
> + return 1;
> + else if (bss->group < new_bss->group)
> + return -1;
> +
> + /* BSS's both have the same group, sort by rank */
> if (bss->rank == new_bss->rank)
> return (bss->signal_strength >
> new_bss->signal_strength) ? 1 : -1;
> @@ -2805,6 +2865,7 @@ static bool station_roam_scan_notify(int err, struct l_queue *bss_list,
> struct handshake_state *hs = netdev_get_handshake(station->netdev);
> struct scan_bss *current_bss = station->connected_bss;
> struct scan_bss *bss;
> + uint32_t cur_bss_group = 0;
> double cur_bss_rank = 0.0;
> static const double RANK_FT_FACTOR = 1.3;
> uint16_t mdid;
> @@ -2835,6 +2896,9 @@ static bool station_roam_scan_notify(int err, struct l_queue *bss_list,
> bss = l_queue_find(bss_list, bss_match_bssid, current_bss->addr);
> if (bss && !station->ap_directed_roaming) {
> cur_bss_rank = bss->rank;
> + cur_bss_group = evaluate_bss_group(current_bss->addr,
> + current_bss->frequency,
> + current_bss->signal_strength);
This is what caused the CI to fail on testPSK-roam. Instead I should
have done the group evaluation on the "bss" pointer, not "current_bss".
This will take into account the most recent scan, rather than the stale
object we have from past scans.
>
> if (hs->mde && bss->mde_present && l_get_le16(bss->mde) == mdid)
> cur_bss_rank *= RANK_FT_FACTOR;
> @@ -2859,6 +2923,9 @@ static bool station_roam_scan_notify(int err, struct l_queue *bss_list,
> while ((bss = l_queue_pop_head(bss_list))) {
> double rank;
> struct roam_bss *rbss;
> + uint32_t group = evaluate_bss_group(bss->addr,
> + bss->frequency,
> + bss->signal_strength);
>
> station_print_scan_bss(bss);
>
> @@ -2889,7 +2956,15 @@ static bool station_roam_scan_notify(int err, struct l_queue *bss_list,
> if (hs->mde && bss->mde_present && l_get_le16(bss->mde) == mdid)
> rank *= RANK_FT_FACTOR;
>
> - if (rank <= cur_bss_rank)
> + /*
> + * First check the group:
> + * - If worse, disregard BSS candidate
> + * - If better, keep BSS candidate
> + * - If equal, compare based on rank
> + */
> + if (group < cur_bss_group)
> + goto next;
> + else if (group == cur_bss_group && rank <= cur_bss_rank)
> goto next;
>
> /*
> @@ -2898,7 +2973,7 @@ static bool station_roam_scan_notify(int err, struct l_queue *bss_list,
> */
> station_update_roam_bss(station, bss);
>
> - rbss = roam_bss_from_scan_bss(bss, rank);
> + rbss = roam_bss_from_scan_bss(bss, rank, group);
>
> l_queue_insert(station->roam_bss_list, rbss,
> roam_bss_rank_compare, NULL);
> @@ -3268,6 +3343,10 @@ static void station_ap_directed_roam(struct station *station,
> l_timeout_remove(station->roam_trigger_timeout);
> station->roam_trigger_timeout = NULL;
>
> + blacklist_add_bss(station->connected_bss->addr,
> + BLACKLIST_REASON_ROAM_REQUESTED);
> + station_debug_event(station, "ap-roam-blacklist-added");
> +
> if (req_mode & WNM_REQUEST_MODE_PREFERRED_CANDIDATE_LIST) {
> l_debug("roam: AP sent a preferred candidate list");
> station_neighbor_report_cb(station->netdev, 0, body + pos,
> @@ -5344,7 +5423,8 @@ static bool station_force_roam_scan_notify(int err, struct l_queue *bss_list,
> /* The various roam routines expect this to be set from scanning */
> station->preparing_roam = true;
> l_queue_push_tail(station->roam_bss_list,
> - roam_bss_from_scan_bss(target, target->rank));
> + roam_bss_from_scan_bss(target, target->rank,
> + 0xffff));
>
> station_update_roam_bss(station, target);
>
next prev parent reply other threads:[~2025-03-26 17:50 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-03-25 18:00 [PATCH v3 01/10] station: always add BSS to network blacklist on failure James Prestwood
2025-03-25 18:00 ` [PATCH v3 02/10] auto-t: add test for disabling the timeout blacklist James Prestwood
2025-03-25 18:00 ` [PATCH v3 03/10] blacklist: include a blacklist reason when adding/finding James Prestwood
2025-03-25 18:00 ` [PATCH v3 04/10] blacklist: fix pruning to remove the entry if its expired James Prestwood
2025-03-25 18:00 ` [PATCH v3 05/10] blacklist: add new blacklist reason, ROAM_REQUESTED James Prestwood
2025-03-25 18:00 ` [PATCH v3 06/10] netdev: add netdev_get_low_signal_thresholds James Prestwood
2025-03-25 18:00 ` [PATCH v3 07/10] station: roam blacklist BSS's, and consider when roaming James Prestwood
2025-03-26 17:49 ` James Prestwood [this message]
2025-03-25 18:00 ` [PATCH v3 08/10] station: roam blacklist AP even mid-roam James Prestwood
2025-03-25 18:00 ` [PATCH v3 09/10] auto-t: add tests for AP roam blacklisting James Prestwood
2025-03-25 18:00 ` [PATCH v3 10/10] doc: document InitialRoamRequestedTimeout James Prestwood
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=d311d92a-b236-4ff1-89bc-377a200e4cb4@gmail.com \
--to=prestwoj@gmail.com \
--cc=iwd@lists.linux.dev \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox