| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <1AD6632B-A69E-406B-A644-440B9C8B929F@xilinx.com>
Date: Mon, 15 Mar 2021 21:42:22 +0000
From: Ben Levinsky <BLEVINSK@...inx.com>
To: Mathieu Poirier <mathieu.poirier@...aro.org>
CC: "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
"linux-remoteproc@...r.kernel.org" <linux-remoteproc@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"linux-arm-kernel@...ts.infradead.org"
<linux-arm-kernel@...ts.infradead.org>,
Michal Simek <michals@...inx.com>,
"Ed T. Mooring" <emooring@...inx.com>
Subject: Re: [PATCH v26 5/5] remoteproc: Add initial zynqmp R5 remoteproc
driver
-----Original Message-----
From: Mathieu Poirier <mathieu.poirier@...aro.org>
Date: Monday, March 15, 2021 at 10:26 AM
To: Ben Levinsky <BLEVINSK@...inx.com>
Cc: "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>, "linux-remoteproc@...r.kernel.org" <linux-remoteproc@...r.kernel.org>, "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>, "linux-arm-kernel@...ts.infradead.org" <linux-arm-kernel@...ts.infradead.org>, Michal Simek <michals@...inx.com>, "Ed T. Mooring" <emooring@...inx.com>
Subject: Re: [PATCH v26 5/5] remoteproc: Add initial zynqmp R5 remoteproc driver
> > +
> > +static void zynqmp_r5_cleanup_mbox(struct zynqmp_r5_rproc *z_rproc)
> > +{
> > + mbox_free_channel(z_rproc->tx_chan);
> > + mbox_free_channel(z_rproc->rx_chan);
> > +}
> > +
> > +/**
> > + * zynqmp_r5_probe - Probes ZynqMP R5 processor device node
> > + * this is called for each individual R5 core to
> > + * set up mailbox, Xilinx platform manager unique ID,
> > + * add to rproc core
>
> The above has changed since last time, which makes it harder for me to
> review your work. From hereon please change only the things I point out so that
> we keep the same goal posts from one revision to the other.
>
> The tabulation needs to be fixed:
>
> * zynqmp_r5_probe - Probes ZynqMP R5 processor device node
> *
> * This is called for each individual R5 core to set up mailbox, Xilinx
> * platform manager unique ID, add to rproc core.
>
> The description is also broken.
>
> [Ben] Ok. How is the following:
> /**
> * zynqmp_r5_probe - Probes ZynqMP R5 processor device node
> *
> * This is called for each individual R5 core to set up mailbox, Xilinx
> * platform manager unique ID, collect SRAM information and wire in
> * driver-specific data to to rproc core.
> *
> * @pdev: domain platform device for current R5 core
> * @node: pointer of the device node for current R5 core
> * @rpu_mode: mode to configure RPU, split or lockstep
> *
> * Return: 0 for success, negative value for failure.
Much better
> */
> static struct zynqmp_r5_rproc *zynqmp_r5_probe(struct platform_device *pdev,
> struct device_node *node,
> enum rpu_oper_mode rpu_mode)
>
>
> > + *
> > + * @pdev: domain platform device for current R5 core
> > + * @node: pointer of the device node for current R5 core
> > + * @rpu_mode: mode to configure RPU, split or lockstep
> > + *
> > + * Return: 0 for success, negative value for failure.
> > + */
> > +static struct zynqmp_r5_rproc *zynqmp_r5_probe(struct platform_device *pdev,
> > + struct device_node *node,
> > + enum rpu_oper_mode rpu_mode)
> > +{
> > + int ret, num_banks;
> > + struct device *dev = &pdev->dev;
> > + struct rproc *rproc_ptr;
> > + struct zynqmp_r5_rproc *z_rproc;
> > + struct device_node *r5_node;
> > +
> > + /* Allocate remoteproc instance */
> > + rproc_ptr = devm_rproc_alloc(dev, dev_name(dev), &zynqmp_r5_rproc_ops,
> > + NULL, sizeof(struct zynqmp_r5_rproc));
> > + if (!rproc_ptr) {
> > + ret = -ENOMEM;
> > + goto error;
> > + }
> > +
> > + rproc_ptr->auto_boot = false;
> > + z_rproc = rproc_ptr->priv;
> > + z_rproc->rproc = rproc_ptr;
> > + r5_node = z_rproc->rproc->dev.parent->of_node;
> > +
> > + /* Set up DMA mask */
> > + ret = dma_set_coherent_mask(dev, DMA_BIT_MASK(32));
> > + if (ret)
> > + goto error;
> > +
> > + /* Get R5 power domain node */
> > + ret = of_property_read_u32(node, "power-domain", &z_rproc->pnode_id);
> > + if (ret)
> > + goto error;
> > +
> > + ret = r5_set_mode(z_rproc, rpu_mode);
> > + if (ret)
> > + goto error;
> > +
> > + if (of_property_read_bool(node, "mboxes")) {
> > + ret = zynqmp_r5_setup_mbox(z_rproc, node);
> > + if (ret)
> > + goto error;
> > + }
> > +
> > + /* go through TCM banks for r5 node */
> > + num_banks = of_count_phandle_with_args(r5_node, BANK_LIST_PROP, NULL);
>
> Shouldn't this be @node instead of @r5_node?
>
> [Ben] Yes this should and will be node.
>
> > + if (num_banks <= 0) {
> > + dev_err(dev, "need to specify TCM banks\n");
> > + ret = -EINVAL;
> > + goto error;
> > + }
> > +
> > + if (num_banks > NUM_SRAMS) {
> > + dev_err(dev, "max number of srams is %d. given: %d \r\n",
> > + NUM_SRAMS, num_banks);
> > + ret = -EINVAL;
> > + goto error;
> > + }
> > +
> > + /* construct collection of srams used by the current R5 core */
> > + for (; num_banks; num_banks--) {
> > + struct resource rsc;
> > + struct device_node *dt_node;
> > + resource_size_t size;
> > + int i;
> > +
> > + dt_node = of_parse_phandle(r5_node, BANK_LIST_PROP, i);
> > + if (!dt_node) {
> > + ret = -EINVAL;
> > + goto error;
> > + }
> > +
> > + ret = of_address_to_resource(dt_node, 0, &rsc);
> > + if (ret < 0) {
> > + of_node_put(dt_node);
> > + goto error;
> > + }
> > +
> > + of_node_put(dt_node);
> > + size = resource_size(&rsc);
> > +
> > + /*
> > + * Find corresponding Xilinx platform management ID.
> > + * The bank information is used in prepare/unprepare and
> > + * parse_fw.
> > + */
> > + for (i = 0; i < NUM_SRAMS; i++) {
> > + if (rsc.start == zynqmp_banks[i].addr) {
> > + z_rproc->srams[i].addr = rsc.start;
> > + z_rproc->srams[i].size = size;
> > + z_rproc->srams[i].id = zynqmp_banks[i].id;
> > + break;
> > + }
> > + }
> > +
> > + if (i == NUM_SRAMS) {
> > + dev_err(dev, "sram %llx is not valid.\n", rsc.start);
> > + ret = -EINVAL;
> > + goto error;
> > + }
> > + }
>
> Everything that is related to the initialisation of srams above should be in a
> function on its own. This too is new code that wasn't requested - the next
> revision needs to include *only* the changes I request. Any improvement on the
> current implementation can be made in future patchsets.
>
>
> [Ben] Makes sense. I will do that going forward. For probe() I will put all the sram information collection functionality in 1 function.
>
> > +
> > + /* Add R5 remoteproc */
> > + ret = devm_rproc_add(dev, rproc_ptr);
> > + if (ret) {
> > + zynqmp_r5_cleanup_mbox(z_rproc);
> > + goto error;
> > + }
> > +
> > + return z_rproc;
> > +error:
> > + return ERR_PTR(ret);
> > +}
> > +
> > +/*
> > + * zynqmp_r5_remoteproc_probe
> > + *
> > + * @pdev: domain platform device for R5 cluster
> > + *
> > + * called when driver is probed, for each R5 core specified in DT,
> > + * setup as needed to do remoteproc-related operations
> > + *
> > + * Return: 0 for success, negative value for failure.
> > + */
> > +static int zynqmp_r5_remoteproc_probe(struct platform_device *pdev)
> > +{
> > + int ret, core_count;
> > + struct device *dev = &pdev->dev;
> > + struct device_node *nc;
> > + enum rpu_oper_mode rpu_mode = PM_RPU_MODE_LOCKSTEP;
> > + struct list_head *cluster; /* list to track each core's rproc */
> > + struct zynqmp_r5_rproc *z_rproc;
> > + struct platform_device *child_pdev;
> > + struct list_head *pos;
> > +
> > + ret = of_property_read_u32(dev->of_node, "xlnx,cluster-mode", &rpu_mode);
> > + if (ret < 0 || (rpu_mode != PM_RPU_MODE_LOCKSTEP &&
> > + rpu_mode != PM_RPU_MODE_SPLIT)) {
> > + dev_err(dev, "invalid cluster mode: ret %d mode %x\n",
> > + ret, rpu_mode);
> > + return ret;
> > + }
> > +
> > + dev_dbg(dev, "RPU configuration: %s\n",
> > + rpu_mode == PM_RPU_MODE_LOCKSTEP ? "lockstep" : "split");
> > +
> > + /*
> > + * if 2 RPUs provided but one is lockstep, then we have an
> > + * invalid configuration.
> > + */
> > +
> > + core_count = of_get_available_child_count(dev->of_node);
> > + if ((rpu_mode == PM_RPU_MODE_LOCKSTEP && core_count != 1) ||
> > + core_count > MAX_RPROCS)
> > + return -EINVAL;
> > +
> > + cluster = devm_kzalloc(dev, sizeof(*cluster), GFP_KERNEL);
> > + if (!cluster)
> > + return -ENOMEM;
> > + INIT_LIST_HEAD(cluster);
> > +
> > + ret = devm_of_platform_populate(dev);
> > + if (ret) {
> > + dev_err(dev, "devm_of_platform_populate failed, ret = %d\n", ret);
> > + return ret;
> > + }
> > +
> > + /* probe each individual r5 core's remoteproc-related info */
> > + for_each_available_child_of_node(dev->of_node, nc) {
> > + child_pdev = of_find_device_by_node(nc);
>
> The device reference needs to be dropped after use, as described in the function
> documentation.
>
> I'm out of time - I will continue tomorrow.
>
> Mathieu
>
>
> [Ben] By this do you mean that for each platform_device should have a call like
> platform_set_drvdata(child_pdev, NULL); if it fails? or something else?
Have another read at the documentation and look at how other people have used
it. You may already be aware but Bootlin's kernel cross-reference tool is
really good for that.
https://elixir.bootlin.com/linux/v5.12-rc3/source
If I understand what you are saying I will add calls for put_device(child_pdev) in error handling and at end of the loop.
Powered by blists - more mailing lists