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
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <e958afe7-b338-47dc-905e-f3223b4f3cdb@collabora.com>
Date: Mon, 30 Jun 2025 11:32:45 +0200
From: AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>
To: Fei Shao <fshao@...omium.org>
Cc: linux-mediatek@...ts.infradead.org, robh@...nel.org, krzk+dt@...nel.org,
 conor+dt@...nel.org, matthias.bgg@...il.com, ulf.hansson@...aro.org,
 y.oudjana@...tonmail.com, wenst@...omium.org, lihongbo22@...wei.com,
 mandyjh.liu@...iatek.com, mbrugger@...e.com, devicetree@...r.kernel.org,
 linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
 linux-pm@...r.kernel.org, kernel@...labora.com
Subject: Re: [PATCH v1 02/13] pmdomain: mediatek: Refactor bus protection
 regmaps retrieval

Il 27/06/25 14:12, Fei Shao ha scritto:
> On Mon, Jun 23, 2025 at 8:02 PM AngeloGioacchino Del Regno
> <angelogioacchino.delregno@...labora.com> wrote:
>>
>> In preparation to add support for new generation SoCs like MT8196,
>> MT6991 and other variants, which require to set bus protection on
>> different busses than the ones found on legacy chips, and to also
>> simplify and reduce memory footprint of this driver, refactor the
>> mechanism to retrieve and use the bus protection regmaps.
>>
>> This is done by removing the three pointers to struct regmap from
>> struct scpsys_domain (allocated for each power domain) and moving
>> them to the main struct scpsys (allocated per driver instance) as
>> an array of pointers to regmap named **bus_prot.
>>
>> That deprecates the old devicetree properties to grab phandles to
>> the three predefined busses (infracfg, infracfg-nao and smi) and
>> replaces it with a new property "mediatek,bus-protection" that is
>> meant to be an array of phandles holding the same busses where
>> required (for now - for legacy SoCs).
>>
>> The new bus protection phandles are indexed by the bus_prot_index
>> member of struct scpsys, used to map "bus type" (ex.: infra, smi,
>> etc) to the specific *bus_prot[x] element.
>>
>> While the old per-power-domain regmap pointers were removed, the
>> support for old devicetree was retained by still checking if the
>> new property (in DT) and new-style declaration (in SoC specific
>> platform data) are both present at probe time.
>>
>> If those are not present, a lookup for the old properties will be
>> done in all of the children of the power controller, and pointers
>> to regmaps will be retrieved with the old properties, but then
>> will be internally remapped to follow the new style regmap anyway
>> as to let this driver benefit of the memory footprint reduction.
>>
>> Finally, it was necessary to change macros in mtk-pm-domains.h and
>> in mt8365-pm-domains.h to make use of the new style bus protection
>> declaration, as the actual HW block is now recognized not by flags
>> but by its own scpsys_bus_prot_block enumeration.
>>
>> The BUS_PROT_(STA)_COMPONENT_{INFRA,INFRA_NAO,SMI} flags were also
>> removed since they are now unused, and because that enumeration was
>> initially meant to vary the logic of bus protection and not the bus
>> where work is performed, anyway!
>>
>> Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>
>> ---
> 
> <snip>
> 
>>
>> +static int scpsys_get_bus_protection_legacy(struct device *dev, struct scpsys *scpsys)
>> +{
>> +       const u8 bp_blocks[3] = {
>> +               BUS_PROT_BLOCK_INFRA, BUS_PROT_BLOCK_SMI, BUS_PROT_BLOCK_INFRA_NAO
>> +       };
>> +       struct device_node *np = dev->of_node;
>> +       struct device_node *node, *smi_np;
>> +       int num_regmaps = 0, i, j;
>> +       struct regmap *regmap[3];
>> +
>> +       /*
>> +        * Legacy code retrieves a maximum of three bus protection handles:
>> +        * some may be optional, or may not be, so the array of bp blocks
>> +        * that is normally passed in as platform data must be dynamically
>> +        * built in this case.
>> +        *
>> +        * Here, try to retrieve all of the regmaps that the legacy code
>> +        * supported and then count the number of the ones that are present,
>> +        * this makes it then possible to allocate the array of bus_prot
>> +        * regmaps and convert all to the new style handling.
>> +        */
>> +       node = of_find_node_with_property(np, "mediatek,infracfg");
>> +       if (node) {
>> +               regmap[0] = syscon_regmap_lookup_by_phandle(node, "mediatek,infracfg");
>> +               of_node_put(node);
>> +               num_regmaps++;
>> +               if (IS_ERR(regmap[0]))
>> +                       return dev_err_probe(dev, PTR_ERR(regmap[0]),
>> +                                            "%pOF: failed to get infracfg regmap\n",
>> +                                            node);
>> +       } else {
>> +               regmap[0] = NULL;
>> +       }
>> +
>> +       node = of_find_node_with_property(np, "mediatek,smi");
>> +       if (node) {
>> +               smi_np = of_parse_phandle(node, "mediatek,smi", 0);
>> +               of_node_put(node);
>> +               if (!smi_np)
>> +                       return -ENODEV;
>> +
>> +               regmap[1] = device_node_to_regmap(smi_np);
>> +               num_regmaps++;
>> +               of_node_put(smi_np);
>> +               if (IS_ERR(regmap[1]))
>> +                       return dev_err_probe(dev, PTR_ERR(regmap[1]),
>> +                                            "%pOF: failed to get SMI regmap\n",
>> +                                            node);
>> +       } else {
>> +               regmap[1] = NULL;
>> +       }
>> +
>> +       node = of_find_node_with_property(np, "mediatek,infracfg-nao");
>> +       if (node) {
>> +               regmap[2] = syscon_regmap_lookup_by_phandle(node, "mediatek,infracfg-nao");
>> +               num_regmaps++;
>> +               of_node_put(node);
>> +               if (IS_ERR(regmap[2]))
>> +                       return dev_err_probe(dev, PTR_ERR(regmap[2]),
>> +                                            "%pOF: failed to get infracfg regmap\n",
>> +                                            node);
>> +       } else {
>> +               regmap[2] = NULL;
>> +       }
>> +
>> +       scpsys->bus_prot = devm_kmalloc_array(dev, num_regmaps,
>> +                                             sizeof(*scpsys->bus_prot), GFP_KERNEL);
>> +       if (!scpsys->bus_prot)
>> +               return -ENOMEM;
>> +
>> +       for (i = 0, j = 0; i < num_regmaps; i++) {
> 
> Did you mean BUS_PROT_BLOCK_COUNT?
> Consider a case where only regmap[2] is configured.
> 

Yep. None of the many platforms that we have tested hit this issue, but it's as
bad as it sounds! :')

Thanks for spotting this one!

Cheers,
Angelo

> Regards,
> Fei
> 
>> +               enum scpsys_bus_prot_block bp_type;
>> +
>> +               if (!regmap[i])
>> +                       continue;
>> +
>> +               bp_type = bp_blocks[i];
>> +               scpsys->bus_prot_index[bp_type] = j;
>> +               scpsys->bus_prot[j] = regmap[i];
>> +
>> +               j++;
>> +       }
>> +
>> +       return 0;
>> +}
>> +
> 
> <snip>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ