Return-path: Received: from wolverine02.qualcomm.com ([199.106.114.251]:63805 "EHLO wolverine02.qualcomm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754371AbcEPTYL (ORCPT ); Mon, 16 May 2016 15:24:11 -0400 From: Maya Erez To: Kalle Valo Cc: Maya Erez , linux-wireless@vger.kernel.org, wil6210@qca.qualcomm.com Subject: [PATCH 3/6] wil6210: protect wil_vring_fini_tx in parallel to tx completions Date: Mon, 16 May 2016 22:23:32 +0300 Message-Id: <1463426615-15523-4-git-send-email-qca_merez@qca.qualcomm.com> (sfid-20160516_212423_128059_992A033C) In-Reply-To: <1463426615-15523-1-git-send-email-qca_merez@qca.qualcomm.com> References: <1463426615-15523-1-git-send-email-qca_merez@qca.qualcomm.com> Sender: linux-wireless-owner@vger.kernel.org List-ID: napi_synchronize is called before releasing the vring, with the assumption that setting txdata->enabled to 0 will prevent handling of this vring in the next scheduled napi. To guarantee this assumption, a memory barrier is added after disabling the txdata. In addition, as the ctx is zeroed in wil_tx_complete after this descriptor is handled (protected by wmb), ctx needs to be checked before releasing this descriptor in wil_vring_free. Signed-off-by: Maya Erez --- drivers/net/wireless/ath/wil6210/txrx.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/drivers/net/wireless/ath/wil6210/txrx.c b/drivers/net/wireless/ath/wil6210/txrx.c index 3909af1..483e063 100644 --- a/drivers/net/wireless/ath/wil6210/txrx.c +++ b/drivers/net/wireless/ath/wil6210/txrx.c @@ -184,6 +184,13 @@ static void wil_vring_free(struct wil6210_priv *wil, struct vring *vring, &vring->va[vring->swtail].tx; ctx = &vring->ctx[vring->swtail]; + if (!ctx) { + wil_dbg_txrx(wil, + "ctx(%d) was already completed\n", + vring->swtail); + vring->swtail = wil_vring_next_tail(vring); + continue; + } *d = *_d; wil_txdesc_unmap(dev, d, ctx); if (ctx->skb) @@ -975,6 +982,13 @@ void wil_vring_fini_tx(struct wil6210_priv *wil, int id) txdata->dot1x_open = false; txdata->enabled = 0; /* no Tx can be in progress or start anew */ spin_unlock_bh(&txdata->lock); + /* napi_synchronize waits for completion of the current NAPI but will + * not prevent the next NAPI run. + * Add a memory barrier to guarantee that txdata->enabled is zeroed + * before napi_synchronize so that the next scheduled NAPI will not + * handle this vring + */ + wmb(); /* make sure NAPI won't touch this vring */ if (test_bit(wil_status_napi_en, wil->status)) napi_synchronize(&wil->napi_tx); -- 1.8.5.2