[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20220408220718.GA2511628@jaehee-ThinkPad-X1-Extreme>
Date: Fri, 8 Apr 2022 18:07:18 -0400
From: Jaehee Park <jhpark1013@...il.com>
To: Stefano Brivio <sbrivio@...hat.com>
Cc: Jérôme Pouiller <jerome.pouiller@...abs.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
linux-staging@...ts.linux.dev, linux-kernel@...r.kernel.org,
outreachy@...ts.linux.dev,
Alison Schofield <alison.schofield@...el.com>
Subject: Re: [PATCH v2] staging: wfx: use container_of() to get vif
On Fri, Apr 08, 2022 at 07:43:12AM +0200, Stefano Brivio wrote:
> Hi Jaehee,
>
> Congrats, almost there! This wasn't necessarily an easy change.
>
Thank you so much for helping me!
> A general comment first: please wait a bit longer between re-postings.
> I received the first patch at 21:50 UTC+2 (my local time), and the
> second one at 5:23 UTC+2.
>
> I could have shared most of the comments below already on your first
> patch, if only I had time to read it. I think something between 12 to
> 15 hours would be appropriate.
>
> Now, to the patch:
>
> On Thu, 7 Apr 2022 23:23:49 -0400
> Jaehee Park <jhpark1013@...il.com> wrote:
>
> > Use container_of() to get vif. This improves the code in two ways:
> > [1] it speeds up the compilation because container_of() saves steps to
> > retrieve vif (the representation of ieee80211_vif), and
>
> "[1] " looks like a reference, perhaps use "- " or "1. " instead.
>
> It would be good to follow the guidelines at:
> https://www.kernel.org/doc/html/latest/process/submitting-patches.html#describe-changes
>
> ...starting from the problem description, even if it's a bit obvious
> here. For example:
>
> Currently, upon virtual interface creation,
> wfx_add_interface() stores a references to the
> corresponding struct ieee80211_vif in private data, for later
> usage. This is not needed as ...
>
> Use container_of() instead. This...
>
Thank you for the improved patch message! What do you think of this:
Currently, upon virtual interface creation, wfx_add_interface() stores
a reference to the corresponding struct ieee80211_vif in private data,
for later usage. This is not needed when using the container_of
construct. This construct already has all the info it needs to retrieve
the reference to the corresponding struct from the offset that is
already available, inherent in container_of(), between its type and
member inputs (struct ieee80211_vif and drv_priv, respectively).
Remove vif (which was previously storing the reference to the struct
ieee80211_vif) from the struct wfx_vif, define a macro
wvif_to_vif(wvif) for container_of(), and replace all wvif->vif with
the newly defined container_of construct.
> > [2] it eliminates the need to define multiple pointers. Previously,
> > after setting the pointer to drv_priv, another pointer had to be
> > defined by vif in struct wfx_vif to point back to ieee80211_vif.
> > Remove vif from the struct wfx_vif, define a macro wvif_to_vif(ptr)
> > for container_of(), and replace all wvif->vif with the newly defined
> > container_of() construct.
> >
> > Signed-off-by: Jaehee Park <jhpark1013@...il.com>
> > ---
> > Note:
> > This is a revised patch to sequence the wfx.h file (with the new defines) to show up first on the diff, which makes the ordering of the diff more logical.
>
> Trim your lines at something between 70 and 76 columns, even if it's
> not part of the commit message -- that will make it easier to reply.
>
> One additional note given where this patch will need to go: for net and
> net-next trees, you're required to describe the change log at the end
> of the commit message itself. That's an exception compared to other
> trees (including staging -- where adding the commit log here is
> appropriate).
>
> > Questions:
> > - These changes had built without errors for me but it would be great if someone can test it on the actual hardware.
> > - We noticed that the development of the wfx driver is happening in the netdev tree instead of Greg's tree. Do you suggest that I use the net-next tree for this change?
> > https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/log/drivers/staging/wfx
> > https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git/log/drivers/staging/wfx
> >
> > drivers/staging/wfx/wfx.h | 3 ++-
> > drivers/staging/wfx/data_rx.c | 4 ++--
> > drivers/staging/wfx/data_tx.c | 2 +-
> > drivers/staging/wfx/key.c | 3 +--
> > drivers/staging/wfx/queue.c | 2 +-
> > drivers/staging/wfx/scan.c | 6 ++---
> > drivers/staging/wfx/sta.c | 45 ++++++++++++++++-------------------
> > 7 files changed, 31 insertions(+), 34 deletions(-)
> >
> > diff --git a/drivers/staging/wfx/wfx.h b/drivers/staging/wfx/wfx.h
> > index 6594cc647c2f..78f2a416fe4f 100644
> > --- a/drivers/staging/wfx/wfx.h
> > +++ b/drivers/staging/wfx/wfx.h
> > @@ -25,6 +25,8 @@
> > #define USEC_PER_TXOP 32 /* see struct ieee80211_tx_queue_params */
> > #define USEC_PER_TU 1024
> >
> > +#define wvif_to_vif(ptr)(container_of((void *)ptr, struct ieee80211_vif, drv_priv))
>
> On top of what Greg said (you're missing a space there) the line is also
> too long, you can split it by doing something like:
>
> #define wvif_to_vif(...) \
> container_of(...)
>
> also:
>
> - parentheses around container_of() are not necessary
>
> - "ptr"? It's obvious that it's a pointer, given the cast. "wvif" would
> be more descriptive
>
> > +
> > struct wfx_hwbus_ops;
> >
> > struct wfx_dev {
> > @@ -61,7 +63,6 @@ struct wfx_dev {
> >
> > struct wfx_vif {
> > struct wfx_dev *wdev;
> > - struct ieee80211_vif *vif;
> > struct ieee80211_channel *channel;
> > int id;
> >
> > diff --git a/drivers/staging/wfx/data_rx.c b/drivers/staging/wfx/data_rx.c
> > index a4b5ffe158e4..98c2223089b8 100644
> > --- a/drivers/staging/wfx/data_rx.c
> > +++ b/drivers/staging/wfx/data_rx.c
> > @@ -24,12 +24,12 @@ static void wfx_rx_handle_ba(struct wfx_vif *wvif, struct ieee80211_mgmt *mgmt)
> > case WLAN_ACTION_ADDBA_REQ:
> > params = le16_to_cpu(mgmt->u.action.u.addba_req.capab);
> > tid = (params & IEEE80211_ADDBA_PARAM_TID_MASK) >> 2;
> > - ieee80211_start_rx_ba_session_offl(wvif->vif, mgmt->sa, tid);
> > + ieee80211_start_rx_ba_session_offl(wvif_to_vif(wvif), mgmt->sa, tid);
>
> Staying under 80 columns is preferred:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=bdc48fa11e46f867ea4d75fa59ee87a7f48be144
>
> but checkpatch.pl won't report this. Perhaps to_vif() would be
> descriptive enough, or you could also just split this line.
>
> Or, even better, especially if you have several usages in a function:
> just grab one reference to vif at the beginning of the function, and
> use that instead.
>
For the revised patch coming up:
I tried different ways to keep the lines below the threshold of 80 columns.
I decided to shorten the macro name from wvif_to_vif to to_vif and for
functions that had more than one instance of vif, define one reference vif
at the beginning and used that instead. I also broke the if-statements that
ran long, into two lines.
In this updated patch there's only 3 lines that exceed 80 by less-than-4
characters. Two of those lines of code could be shorted but it involved
defining two more variables, and I didn't think it was worth it because
it could potentially make the code less readable.
> > break;
> > case WLAN_ACTION_DELBA:
> > params = le16_to_cpu(mgmt->u.action.u.delba.params);
> > tid = (params & IEEE80211_DELBA_PARAM_TID_MASK) >> 12;
> > - ieee80211_stop_rx_ba_session_offl(wvif->vif, mgmt->sa, tid);
> > + ieee80211_stop_rx_ba_session_offl(wvif_to_vif(wvif), mgmt->sa, tid);
>
> Same here.
>
> > break;
> > }
> > }
> > diff --git a/drivers/staging/wfx/data_tx.c b/drivers/staging/wfx/data_tx.c
> > index e07381b2ff4d..fd79e03a08c5 100644
> > --- a/drivers/staging/wfx/data_tx.c
> > +++ b/drivers/staging/wfx/data_tx.c
> > @@ -216,7 +216,7 @@ static u8 wfx_tx_get_link_id(struct wfx_vif *wvif, struct ieee80211_sta *sta,
> >
> > if (sta_priv && sta_priv->link_id)
> > return sta_priv->link_id;
> > - if (wvif->vif->type != NL80211_IFTYPE_AP)
> > + if (wvif_to_vif(wvif)->type != NL80211_IFTYPE_AP)
> > return 0;
> > if (is_multicast_ether_addr(da))
> > return 0;
> > diff --git a/drivers/staging/wfx/key.c b/drivers/staging/wfx/key.c
> > index 8f23e8d42bd4..71b1fca70248 100644
> > --- a/drivers/staging/wfx/key.c
> > +++ b/drivers/staging/wfx/key.c
> > @@ -174,7 +174,7 @@ static int wfx_add_key(struct wfx_vif *wvif, struct ieee80211_sta *sta,
> > k.type = fill_tkip_pair(&k.key.tkip_pairwise_key, key, sta->addr);
> > else
> > k.type = fill_tkip_group(&k.key.tkip_group_key, key, &seq,
> > - wvif->vif->type);
> > + wvif_to_vif(wvif)->type);
> > } else if (key->cipher == WLAN_CIPHER_SUITE_CCMP) {
> > if (pairwise)
> > k.type = fill_ccmp_pair(&k.key.aes_pairwise_key, key, sta->addr);
> > @@ -224,4 +224,3 @@ int wfx_set_key(struct ieee80211_hw *hw, enum set_key_cmd cmd, struct ieee80211_
> > mutex_unlock(&wvif->wdev->conf_mutex);
> > return ret;
> > }
> > -
> > diff --git a/drivers/staging/wfx/queue.c b/drivers/staging/wfx/queue.c
> > index 729825230db2..a2745d1024c6 100644
> > --- a/drivers/staging/wfx/queue.c
> > +++ b/drivers/staging/wfx/queue.c
> > @@ -207,7 +207,7 @@ bool wfx_tx_queues_has_cab(struct wfx_vif *wvif)
> > {
> > int i;
> >
> > - if (wvif->vif->type != NL80211_IFTYPE_AP)
> > + if (wvif_to_vif(wvif)->type != NL80211_IFTYPE_AP)
> > return false;
> > for (i = 0; i < IEEE80211_NUM_ACS; ++i)
> > /* Note: since only AP can have mcast frames in queue and only one vif can be AP,
> > diff --git a/drivers/staging/wfx/scan.c b/drivers/staging/wfx/scan.c
> > index 7f34f0d322f9..5c25fde3fc41 100644
> > --- a/drivers/staging/wfx/scan.c
> > +++ b/drivers/staging/wfx/scan.c
> > @@ -25,7 +25,7 @@ static int update_probe_tmpl(struct wfx_vif *wvif, struct cfg80211_scan_request
> > {
> > struct sk_buff *skb;
> >
> > - skb = ieee80211_probereq_get(wvif->wdev->hw, wvif->vif->addr, NULL, 0, req->ie_len);
> > + skb = ieee80211_probereq_get(wvif->wdev->hw, wvif_to_vif(wvif)->addr, NULL, 0, req->ie_len);
>
> ...and here,
>
> > if (!skb)
> > return -ENOMEM;
> >
> > @@ -75,8 +75,8 @@ static int send_scan_req(struct wfx_vif *wvif, struct cfg80211_scan_request *req
> > } else {
> > ret = wvif->scan_nb_chan_done;
> > }
> > - if (req->channels[start_idx]->max_power != wvif->vif->bss_conf.txpower)
> > - wfx_hif_set_output_power(wvif, wvif->vif->bss_conf.txpower);
> > + if (req->channels[start_idx]->max_power != wvif_to_vif(wvif)->bss_conf.txpower)
> > + wfx_hif_set_output_power(wvif, wvif_to_vif(wvif)->bss_conf.txpower);
>
>
> > wfx_tx_unlock(wvif->wdev);
> > return ret;
> > }
> > diff --git a/drivers/staging/wfx/sta.c b/drivers/staging/wfx/sta.c
> > index 03025ef7f1be..b634c8e004db 100644
> > --- a/drivers/staging/wfx/sta.c
> > +++ b/drivers/staging/wfx/sta.c
> > @@ -132,7 +132,7 @@ void wfx_configure_filter(struct ieee80211_hw *hw, unsigned int changed_flags,
> > filter_bssid = true;
> >
> > /* In AP mode, chip can reply to probe request itself */
> > - if (*total_flags & FIF_PROBE_REQ && wvif->vif->type == NL80211_IFTYPE_AP) {
> > + if (*total_flags & FIF_PROBE_REQ && wvif_to_vif(wvif)->type == NL80211_IFTYPE_AP) {
>
> and here,
>
> > dev_dbg(wdev->dev, "do not forward probe request in AP mode\n");
> > *total_flags &= ~FIF_PROBE_REQ;
> > }
> > @@ -153,18 +153,18 @@ static int wfx_get_ps_timeout(struct wfx_vif *wvif, bool *enable_ps)
> > struct ieee80211_channel *chan0 = NULL, *chan1 = NULL;
> > struct ieee80211_conf *conf = &wvif->wdev->hw->conf;
> >
> > - WARN(!wvif->vif->bss_conf.assoc && enable_ps,
> > + WARN(!wvif_to_vif(wvif)->bss_conf.assoc && enable_ps,
> > "enable_ps is reliable only if associated");
> > if (wdev_to_wvif(wvif->wdev, 0))
> > - chan0 = wdev_to_wvif(wvif->wdev, 0)->vif->bss_conf.chandef.chan;
> > + chan0 = wvif_to_vif(wdev_to_wvif(wvif->wdev, 0))->bss_conf.chandef.chan;
> > if (wdev_to_wvif(wvif->wdev, 1))
> > - chan1 = wdev_to_wvif(wvif->wdev, 1)->vif->bss_conf.chandef.chan;
> > - if (chan0 && chan1 && wvif->vif->type != NL80211_IFTYPE_AP) {
> > + chan1 = wvif_to_vif(wdev_to_wvif(wvif->wdev, 1))->bss_conf.chandef.chan;
> > + if (chan0 && chan1 && wvif_to_vif(wvif)->type != NL80211_IFTYPE_AP) {
> > if (chan0->hw_value == chan1->hw_value) {
> > /* It is useless to enable PS if channels are the same. */
> > if (enable_ps)
> > *enable_ps = false;
> > - if (wvif->vif->bss_conf.assoc && wvif->vif->bss_conf.ps)
> > + if (wvif_to_vif(wvif)->bss_conf.assoc && wvif_to_vif(wvif)->bss_conf.ps)
> > dev_info(wvif->wdev->dev, "ignoring requested PS mode");
> > return -1;
> > }
> > @@ -177,8 +177,8 @@ static int wfx_get_ps_timeout(struct wfx_vif *wvif, bool *enable_ps)
> > return 30;
> > }
> > if (enable_ps)
> > - *enable_ps = wvif->vif->bss_conf.ps;
> > - if (wvif->vif->bss_conf.assoc && wvif->vif->bss_conf.ps)
> > + *enable_ps = wvif_to_vif(wvif)->bss_conf.ps;
> > + if (wvif_to_vif(wvif)->bss_conf.assoc && wvif_to_vif(wvif)->bss_conf.ps)
> > return conf->dynamic_ps_timeout;
> > else
> > return -1;
> > @@ -189,7 +189,7 @@ int wfx_update_pm(struct wfx_vif *wvif)
> > int ps_timeout;
> > bool ps;
> >
> > - if (!wvif->vif->bss_conf.assoc)
> > + if (!wvif_to_vif(wvif)->bss_conf.assoc)
> > return 0;
> > ps_timeout = wfx_get_ps_timeout(wvif, &ps);
> > if (!ps)
> > @@ -215,7 +215,7 @@ int wfx_conf_tx(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
> > mutex_lock(&wdev->conf_mutex);
> > assign_bit(queue, &wvif->uapsd_mask, params->uapsd);
> > wfx_hif_set_edca_queue_params(wvif, queue, params);
> > - if (wvif->vif->type == NL80211_IFTYPE_STATION && old_uapsd != wvif->uapsd_mask) {
> > + if (wvif_to_vif(wvif)->type == NL80211_IFTYPE_STATION && old_uapsd != wvif->uapsd_mask) {
> > wfx_hif_set_uapsd_info(wvif, wvif->uapsd_mask);
> > wfx_update_pm(wvif);
> > }
> > @@ -242,20 +242,20 @@ void wfx_event_report_rssi(struct wfx_vif *wvif, u8 raw_rcpi_rssi)
> > int cqm_evt;
> >
> > rcpi_rssi = raw_rcpi_rssi / 2 - 110;
> > - if (rcpi_rssi <= wvif->vif->bss_conf.cqm_rssi_thold)
> > + if (rcpi_rssi <= wvif_to_vif(wvif)->bss_conf.cqm_rssi_thold)
> > cqm_evt = NL80211_CQM_RSSI_THRESHOLD_EVENT_LOW;
> > else
> > cqm_evt = NL80211_CQM_RSSI_THRESHOLD_EVENT_HIGH;
> > - ieee80211_cqm_rssi_notify(wvif->vif, cqm_evt, rcpi_rssi, GFP_KERNEL);
> > + ieee80211_cqm_rssi_notify(wvif_to_vif(wvif), cqm_evt, rcpi_rssi, GFP_KERNEL);
> > }
> >
> > static void wfx_beacon_loss_work(struct work_struct *work)
> > {
> > struct wfx_vif *wvif = container_of(to_delayed_work(work), struct wfx_vif,
> > beacon_loss_work);
> > - struct ieee80211_bss_conf *bss_conf = &wvif->vif->bss_conf;
> > + struct ieee80211_bss_conf *bss_conf = &wvif_to_vif(wvif)->bss_conf;
> >
> > - ieee80211_beacon_loss(wvif->vif);
> > + ieee80211_beacon_loss(wvif_to_vif(wvif));
> > schedule_delayed_work(to_delayed_work(work), msecs_to_jiffies(bss_conf->beacon_int));
> > }
> >
> > @@ -323,13 +323,13 @@ static int wfx_upload_ap_templates(struct wfx_vif *wvif)
> > {
> > struct sk_buff *skb;
> >
> > - skb = ieee80211_beacon_get(wvif->wdev->hw, wvif->vif);
> > + skb = ieee80211_beacon_get(wvif->wdev->hw, wvif_to_vif(wvif));
> > if (!skb)
> > return -ENOMEM;
> > wfx_hif_set_template_frame(wvif, skb, HIF_TMPLT_BCN, API_RATE_INDEX_B_1MBPS);
> > dev_kfree_skb(skb);
> >
> > - skb = ieee80211_proberesp_get(wvif->wdev->hw, wvif->vif);
> > + skb = ieee80211_proberesp_get(wvif->wdev->hw, wvif_to_vif(wvif));
> > if (!skb)
> > return -ENOMEM;
> > wfx_hif_set_template_frame(wvif, skb, HIF_TMPLT_PRBRES, API_RATE_INDEX_B_1MBPS);
> > @@ -339,7 +339,7 @@ static int wfx_upload_ap_templates(struct wfx_vif *wvif)
> >
> > static void wfx_set_mfp_ap(struct wfx_vif *wvif)
> > {
> > - struct sk_buff *skb = ieee80211_beacon_get(wvif->wdev->hw, wvif->vif);
> > + struct sk_buff *skb = ieee80211_beacon_get(wvif->wdev->hw, wvif_to_vif(wvif));
> > const int ieoffset = offsetof(struct ieee80211_mgmt, u.beacon.variable);
> > const u16 *ptr = (u16 *)cfg80211_find_ie(WLAN_EID_RSN, skb->data + ieoffset,
> > skb->len - ieoffset);
> > @@ -389,7 +389,7 @@ void wfx_stop_ap(struct ieee80211_hw *hw, struct ieee80211_vif *vif)
> > static void wfx_join(struct wfx_vif *wvif)
> > {
> > int ret;
> > - struct ieee80211_bss_conf *conf = &wvif->vif->bss_conf;
> > + struct ieee80211_bss_conf *conf = &wvif_to_vif(wvif)->bss_conf;
> > struct cfg80211_bss *bss = NULL;
> > u8 ssid[IEEE80211_MAX_SSID_LEN];
> > const u8 *ssid_ie = NULL;
> > @@ -420,7 +420,7 @@ static void wfx_join(struct wfx_vif *wvif)
> > wvif->join_in_progress = true;
> > ret = wfx_hif_join(wvif, conf, wvif->channel, ssid, ssid_len);
> > if (ret) {
> > - ieee80211_connection_loss(wvif->vif);
> > + ieee80211_connection_loss(wvif_to_vif(wvif));
> > wfx_reset(wvif);
> > } else {
> > /* Due to beacon filtering it is possible that the AP's beacon is not known for the
> > @@ -440,7 +440,7 @@ static void wfx_join_finalize(struct wfx_vif *wvif, struct ieee80211_bss_conf *i
> >
> > rcu_read_lock(); /* protect sta */
> > if (info->bssid && !info->ibss_joined)
> > - sta = ieee80211_find_sta(wvif->vif, info->bssid);
> > + sta = ieee80211_find_sta(wvif_to_vif(wvif), info->bssid);
> > if (sta && sta->ht_cap.ht_supported)
> > ampdu_density = sta->ht_cap.ampdu_density;
> > if (sta && sta->ht_cap.ht_supported &&
> > @@ -565,7 +565,7 @@ static int wfx_update_tim(struct wfx_vif *wvif)
> > u16 tim_offset, tim_length;
> > u8 *tim_ptr;
> >
> > - skb = ieee80211_beacon_get_tim(wvif->wdev->hw, wvif->vif, &tim_offset, &tim_length);
> > + skb = ieee80211_beacon_get_tim(wvif->wdev->hw, wvif_to_vif(wvif), &tim_offset, &tim_length);
> > if (!skb)
> > return -ENOENT;
> > tim_ptr = skb->data + tim_offset;
> > @@ -707,8 +707,6 @@ int wfx_add_interface(struct ieee80211_hw *hw, struct ieee80211_vif *vif)
> > return -EOPNOTSUPP;
> > }
> >
> > - /* FIXME: prefer use of container_of() to get vif */
> > - wvif->vif = vif;
> > wvif->wdev = wdev;
> >
> > wvif->link_id_map = 1; /* link-id 0 is reserved for multicast */
> > @@ -767,7 +765,6 @@ void wfx_remove_interface(struct ieee80211_hw *hw, struct ieee80211_vif *vif)
> >
> > cancel_delayed_work_sync(&wvif->beacon_loss_work);
> > wdev->vif[wvif->id] = NULL;
> > - wvif->vif = NULL;
> >
> > mutex_unlock(&wdev->conf_mutex);
> >
>
> The rest looks good to me, except that, as Alison and Greg pointed out,
> this isn't in staging anymore:
> https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git/commit/?h=staging-next&id=4a5fb1bbcdf1cccae1f6b9c0277b3796b2a468ef
>
Yes sorry Alison -- I didn't understand initially. I got it now!
> -- even if it was when we first discussed this change, and also,
> as you already mentioned, you'll need to find somebody to test this on
> actual hardware.
>
> --
> Stefano
>
I will take the updated patch to the wireless-next tree. I have one ready
to send but before I do that I should probably confirm and test with
someone that has the hardware (maybe Jérôme?) before I send the patch in.
Meanwhile, I'll send a revised patch v3 to staging.
Thanks!
Jaehee
Powered by blists - more mailing lists