From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (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 95E5F40B6F4 for ; Sat, 8 Aug 2026 13:13:54 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1786194835; cv=none; b=XxI0iXwdIis2QcWd4Jk0rfaXTVtMUuuIFN3dmYfPM/AlcvFNSDlefLsqrbeDZODk3+P00UA0xHnoYIMnAq5QVWBQnyD3bs+q1skqDZ719K5ZdDgnWRRh6za1pdtjld31vqXmL4n8yYZ6nvSYI2vyjOzecBQNF2lEW1ADWAODkJQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1786194835; c=relaxed/simple; bh=PBFE/PHdHO2myD67NzWHwdIOykuIWR7TtCC1z+SF3zk=; h=From:Subject:To:Cc:In-Reply-To:References:Content-Type:Date: Message-Id; b=NBUMVX+jDSvGIRaK9ZhzIY4+G0TYDqYGViy2bbGAfQr3jBMx10FF4FKg/tkz/0UaCw8r4iXKBZ/YlbfcrDe46C8CTc/oWLfx2xAl+fVdMvoDih4cyru2GWrjlX64Y9GcTtdW+wXiKwROUBRFI/VE3SIcxwgXiL26WIaWn3X8yOE= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=VqPLUKR0; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="VqPLUKR0" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 3C1311F00A3A; Sat, 8 Aug 2026 13:13:54 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1786194834; bh=2W/CeBBlW5Vy8L4jXS3tUysbzxe8fYDl/XD4ibvzd+A=; h=From:Subject:Reply-To:To:Cc:In-Reply-To:References:Date; b=VqPLUKR0E2LNXuJo+cd5ANOgBKbSHrnoHLeXbs0hwQOJ9Hs382r3/rKlMOtuADyNo BungRIWwpwdiIpYmIEBCcobw2s1x6PRlfuSovxbPDC6dv1fpZygVMh/Wh9FlmP0yI5 pX+9Xg5QPYmiFB/89xuwMbmrlP6FfyBK/mg8D5qmez7jZojbxCj6wpaIKjaP6rksVI m4gZbllkY44m5c41r+F9XQo9Ij+L/yejNjCjS/0x9DkTAWYjovKqBIqiAULAHeCphZ 1hg4c8IH5N1EJDL/O+IcmGIqUj3do62Zy+8rZvsZKMbUBVtXzUkoDS3VyURIIfObqa Vfwwfl64S1zTg== From: sashiko-bot@kernel.org Subject: Re: [PATCH net-next v11 06/12] net: phylink: support late PCS provider attach Reply-To: sashiko-reviews@lists.linux.dev To: "Christian Marangi" Cc: conor+dt@kernel.org, robh@kernel.org, devicetree@vger.kernel.org In-Reply-To: <20260807131301.192060-7-ansuelsmth@gmail.com> References: <20260807131301.192060-1-ansuelsmth@gmail.com> <20260807131301.192060-7-ansuelsmth@gmail.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Sat, 08 Aug 2026 13:13:53 +0000 Message-Id: <20260808131354.3C1311F00A3A@smtp.kernel.org> Precedence: bulk X-Mailing-List: devicetree@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Thank you for your contribution! Sashiko AI review found 4 potential issue(= s) to consider: - [Critical] Self-deadlock in fwnode_pcs_del_provider via FWNODE_PCS_PROVID= ER_DEL notification. - [High] Error paths in phylink_create skip necessary cleanup, leading to U= se-After-Free of the phylink instance and dangling PCS pointers. - [High] phylink_del_pcs unconditionally calls list_del, crashing on PCS st= ructures not tracked in the phylink list. - [High] Missing reference counting on struct phylink_pcs pointers allows a= TOCTOU Use-After-Free during asynchronous provider removal. -- --- Patch [6]: [PATCH net-next v11 06/12] net: phylink: support late PCS pr= ovider attach --- commit d404b5bc48ec98384e8bad4f7b0b1af967121eee Author: Christian Marangi net: phylink: support late PCS provider attach This patch adds support for attaching a PCS provider to a phylink instance after the instance has been created and started. It introduces FWNODE_PCS_PROVIDER_ADD to notify phylink when a new PCS provider becomes available, linking it and triggering a resolve step to recover links. > diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c > index 548e783ae2aaa..69d4f6675db0f 100644 > --- a/drivers/net/phy/phylink.c > +++ b/drivers/net/phy/phylink.c [ ... ] > @@ -1939,7 +1960,7 @@ static int phylink_fill_available_pcs(struct phylin= k *pl, > if (!pcs) > continue; > =20 > - list_add_tail(&pcs->list, &pl->pcs_list); > + phylink_add_pcs(pl, pcs); [Severity: High] Could the lack of reference counting on the returned phylink_pcs pointers lead to a use-after-free here? fwnode_pcs_get() and its underlying provider callbacks return a pointer without taking a reference. If a concurrent provider unbind occurs right before pl->state_mutex is acquired, the provider could be removed and the memory freed. When the lock is finally acquired in phylink_fill_available_pcs(), could this add a freed pointer to the list? > } > =20 > mutex_unlock(&pl->state_mutex); > @@ -1991,7 +2012,19 @@ static int pcs_provider_notify(struct notifier_blo= ck *self, > =20 > mutex_lock(&pl->state_mutex); > =20 > - phylink_del_pcs(pl, pcs); > + switch (val) { > + case FWNODE_PCS_PROVIDER_ADD: > + phylink_add_pcs(pl, pcs); > + > + /* Force an interface reconfig if major config fail */ > + if (pl->major_config_failed) > + pl->force_major_config =3D true; > + > + break; > + case FWNODE_PCS_PROVIDER_DEL: > + phylink_del_pcs(pl, pcs); [Severity: High] Will this cause a crash if the PCS being removed was never actually added to the list? When handling a provider removal, phylink_del_pcs() is called unconditionally based purely on its presence in the device tree. If the driver's fill_available_pcs() omitted it, or if phylink_create() failed before it could be added, the pcs->list remains uninitialized. Would this lead to list_del(&pcs->list) dereferencing a NULL pointer? [Severity: Critical] Could this notifier chain lead to a self-deadlock when removing a provider? When fwnode_pcs_del_provider() is called, it acquires fwnode_pcs_mutex and fires the FWNODE_PCS_PROVIDER_DEL notification. This notifier callback, pcs_provider_notify(), responds by calling fwnode_phylink_pcs_get_from_fwnode(): pcs_provider_notify() pcs =3D fwnode_phylink_pcs_get_from_fwnode(pl_fwnode, pcs_fwnode); This subsequently calls fwnode_pcs_get_from_pcsspec(), which attempts to acquire the exact same fwnode_pcs_mutex lock. Since the lock is not recursive, will this result in a deadlock on the thread removing the provider? > + break; > + } > =20 > /* Refresh supported interfaces */ > phy_interface_copy(pl->supported_interfaces, > @@ -2077,10 +2110,6 @@ struct phylink *phylink_create(struct phylink_conf= ig *config, > =20 > mutex_lock(&pl->state_mutex); > =20 > - /* Link available PCS to phylink */ > - list_for_each_entry(pcs, &pl->pcs_list, list) > - pcs->phylink =3D pl; > - [Severity: High] Does the error handling in phylink_create() leak dangling pointers and use freed memory if a failure occurs after the notifier is registered? The PCS notifier is registered early in phylink_create(). If an asynchronous FWNODE_PCS_PROVIDER_ADD event fires concurrently, pcs_provider_notify() can queue the pl->resolve work item. If phylink_create() subsequently fails (for example, in phylink_fill_available_pcs()), the error path jumps to unregister_pcs_notify and frees the phylink structure: unregister_pcs_notify: unregister_fwnode_pcs_notifier(&pl->fwnode_pcs_nb); free_pl: kfree(pl); This path does not call cancel_work_sync(&pl->resolve), leaving the queued work to execute on freed memory. Additionally, because the error path skips unlinking the PCS list, any PCS dynamically added by the notifier will be left with a dangling pcs->phylink pointer. --=20 Sashiko AI review =C2=B7 https://sashiko.dev/#/patchset/20260807131301.1920= 60-1-ansuelsmth@gmail.com?part=3D6