From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 74D4A14A81 for ; Mon, 23 Oct 2023 11:05:15 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linuxfoundation.org header.i=@linuxfoundation.org header.b="kcq0O5Eo" Received: by smtp.kernel.org (Postfix) with ESMTPSA id E5FB0C433C9; Mon, 23 Oct 2023 11:05:14 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1698059115; bh=ly+FwnJsXK6aE7n0TagRNVBdQPqpMCoVSJrptp6pbiQ=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=kcq0O5EoZe3J715UhTyzbzXIeQFEUP1uA8d8JPrdbU873cOCnwiGFe/WLY3KkhoxC xepp7qImfRU99KrW734G3BGfGtIlm6hCqV+VefnJiCv+Z9Y6QkDN72Fn6HAfO5VaHd Sn+gFOAE0BuDyNxmsiOeFggI5qTqaj2WRdsRxwtI= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Shailend Chand , Paolo Abeni Subject: [PATCH 6.5 069/241] gve: Do not fully free QPL pages on prefill errors Date: Mon, 23 Oct 2023 12:54:15 +0200 Message-ID: <20231023104835.581440837@linuxfoundation.org> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20231023104833.832874523@linuxfoundation.org> References: <20231023104833.832874523@linuxfoundation.org> User-Agent: quilt/0.67 X-stable: review X-Patchwork-Hint: ignore Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit 6.5-stable review patch. If anyone has any objections, please let me know. ------------------ From: Shailend Chand commit 95535e37e8959f50e7aee365a5bdc9e5ed720443 upstream. The prefill function should have only removed the page count bias it added. Fully freeing the page will cause gve_free_queue_page_list to free a page the driver no longer owns. Fixes: 82fd151d38d9 ("gve: Reduce alloc and copy costs in the GQ rx path") Signed-off-by: Shailend Chand Link: https://lore.kernel.org/r/20231014014121.2843922-1-shailend@google.com Signed-off-by: Paolo Abeni Signed-off-by: Greg Kroah-Hartman --- drivers/net/ethernet/google/gve/gve_rx.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/google/gve/gve_rx.c b/drivers/net/ethernet/google/gve/gve_rx.c index d1da7413dc4d..e84a066aa1a4 100644 --- a/drivers/net/ethernet/google/gve/gve_rx.c +++ b/drivers/net/ethernet/google/gve/gve_rx.c @@ -146,7 +146,7 @@ static int gve_prefill_rx_pages(struct gve_rx_ring *rx) err = gve_rx_alloc_buffer(priv, &priv->pdev->dev, &rx->data.page_info[i], &rx->data.data_ring[i]); if (err) - goto alloc_err; + goto alloc_err_rda; } if (!rx->data.raw_addressing) { @@ -171,12 +171,26 @@ static int gve_prefill_rx_pages(struct gve_rx_ring *rx) return slots; alloc_err_qpl: + /* Fully free the copy pool pages. */ while (j--) { page_ref_sub(rx->qpl_copy_pool[j].page, rx->qpl_copy_pool[j].pagecnt_bias - 1); put_page(rx->qpl_copy_pool[j].page); } -alloc_err: + + /* Do not fully free QPL pages - only remove the bias added in this + * function with gve_setup_rx_buffer. + */ + while (i--) + page_ref_sub(rx->data.page_info[i].page, + rx->data.page_info[i].pagecnt_bias - 1); + + gve_unassign_qpl(priv, rx->data.qpl->id); + rx->data.qpl = NULL; + + return err; + +alloc_err_rda: while (i--) gve_rx_free_buffer(&priv->pdev->dev, &rx->data.page_info[i], -- 2.42.0