linux-kernel - Re: [PATCH v4] ata/pata_buddha: Probe via modalias instead of initcall

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: <d9fa8aca-62a4-5d4a-b63f-bdd628e6b304@enpas.org>
Date:   Sun, 11 Aug 2019 21:28:33 +0200
From:   Max Staudt <max@...as.org>
To:     b.zolnierkie@...sung.com, axboe@...nel.dk
Cc:     linux-ide@...r.kernel.org, linux-m68k@...r.kernel.org,
        linux-kernel@...r.kernel.org, glaubitz@...sik.fu-berlin.de,
        schmitzmic@...il.com, geert@...ux-m68k.org
Subject: Re: [PATCH v4] ata/pata_buddha: Probe via modalias instead of
 initcall

Replying to my own patch with two more questions:

On 08/11/2019 05:36 PM, Max Staudt wrote:
> -		/* allocate host */
> -		host = ata_host_alloc(&z->dev, nr_ports);

Actually, this is an issue even the existing pata_buddha has: ata_host_alloc() will dev_set_drvdata(dev, host) which is fine on Buddha and Catweasel, but conflicts with zorro8390's own dev_set_drvdata() on an X-Surf board. Thus, if both pata_buddha and zorro8390 are active, only one can be unloaded. The original ide/buddha driver does not have this problem as far as I can see.

This should be resolved once we get around to MFD support, as Geert suggested.

Shall we leave this as-is, as it's not really a change from the status quo in pata_buddha?

> +static int __init pata_buddha_late_init(void)
> +{
> +        struct zorro_dev *z = NULL;
> +
> +	pr_info("pata_buddha: Scanning for stand-alone IDE controllers...\n");
> +	zorro_register_driver(&pata_buddha_driver);
> +
> +	pr_info("pata_buddha: Scanning for X-Surf boards...\n");
> +        while ((z = zorro_find_device(ZORRO_PROD_INDIVIDUAL_COMPUTERS_X_SURF, z))) {
> +		static struct zorro_device_id xsurf_ent =
> +			{ ZORRO_PROD_INDIVIDUAL_COMPUTERS_X_SURF, BOARD_XSURF};
> +
> +		pata_buddha_probe(z, &xsurf_ent);
> +        }
> +
> +        return 0;
> +}

This is suboptimal, as we don't release memory in case pata_buddha_probe() fails. Any suggestions?

> +static void __exit pata_buddha_exit(void)
> +{
> +	struct zorro_dev *z = NULL;
> +
> +	pr_info("pata_buddha: Releasing X-Surf boards...\n");
> +        while ((z = zorro_find_device(ZORRO_PROD_INDIVIDUAL_COMPUTERS_X_SURF, z))) {
> +		struct ata_host *host = dev_get_drvdata(&z->dev);
> +
> +		if (host)
> +			ata_host_detach(host);
> +        }

I guess that here we also need to manually release the resources we allocated with devm_* above. Any ideas?

Thanks

Max