From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from magratgarlick.emantor.de ([2a01:4f8:c17:c88::2] helo=margratgarlick.emantor.de) by bombadil.infradead.org with esmtps (Exim 4.87 #1 (Red Hat Linux)) id 1e41M5-0000q6-95 for ath10k@lists.infradead.org; Mon, 16 Oct 2017 09:06:40 +0000 References: <5a4860eb-3734-7587-d81f-9c3de6ff11c8@leemhuis.info> <46f918c5-b07d-e397-2f3d-8136c7c1a8f3@qti.qualcomm.com> From: Rouven Czerwinski Subject: Re: ath10k-regression due to "ath10k: fix napi_poll budget overflow" c9353bf483d3 In-Reply-To: Date: Mon, 16 Oct 2017 11:04:51 +0200 Message-ID: <87zi8rv4y4.fsf@czerwinskis.de> MIME-Version: 1.0 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: "ath10k" Errors-To: ath10k-bounces+kvalo=adurom.com@lists.infradead.org To: ath10k@lists.infradead.org Cc: Ryan Hsu , Thorsten Leemhuis Hello, here is my analysis of the problem: Commit c9353bf483d3 plumps the budget down into the ath10k_htt_rx_extract_amsdu function. Per extracted msdu frame the budget is reduced by one until either we have no budget left or all msdus are extracted from the amsdu. The function then checks whether all frames were extracted, if not it returns -EAGAIN. The error happens if we have to extract more frames then we have budget_left, because not all frames are extracted and ath10k_htt_rx_in_ord_ind enters this error path: case -EAGAIN: /* fall through */ default: /* Should not happen. */ ath10k_warn(ar, "failed to extract amsdu: %d\n", ret); htt->rx_confused = true; __skb_queue_purge(&list); return -EIO; Which is not recoverable by the driver, only through hardware reset. MfG Rouven Czerwinski _______________________________________________ ath10k mailing list ath10k@lists.infradead.org http://lists.infradead.org/mailman/listinfo/ath10k