From: Ondrej Zary <linux@rainbow-software.org>
To: Finn Thain <fthain@telegraphics.com.au>
Cc: Christoph Hellwig <hch@infradead.org>,
linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/3] g_NCR5380: Stop using scsi_module.c
Date: Tue, 27 Sep 2016 14:55:17 +0200 [thread overview]
Message-ID: <201609271455.17930.linux@rainbow-software.org> (raw)
In-Reply-To: <alpine.LNX.2.00.1609261132001.4881@nippy.intranet>
On Monday 26 September 2016, Finn Thain wrote:
[...]
> Instead of this:
> > + if (scsi_add_host(instance, pdev)) {
> > + free_irq(irq, instance);
> > + goto out_unregister;
> > + }
> > + scsi_scan_host(instance);
> > + return instance;
>
> please use the following (note the missing NCR5380_exit() call):
>
> + if (scsi_add_host(instance, pdev))
> + goto out_free_irq;
> + scsi_scan_host(instance);
> + return(instance);
> +
> +out_free_irq:
> + if (instance->irq != NO_IRQ)
> + free_irq(instance->irq, instance);
> + NCR5380_exit(instance);
>
> > out_unregister:
> > - scsi_unregister(instance);
> > + scsi_host_put(instance);
> > out_release:
> > #ifndef SCSI_G_NCR5380_MEM
> > - release_region(card.NCR5380_map_name, region_size);
> > + release_region(base, region_size);
> > #else
> > iounmap(iomem);
> > release_mem_region(base, iomem_size);
> > #endif
> > - return 0;
> > + return NULL;
> > }
> >
> > -/**
> > - * generic_NCR5380_release_resources - free resources
> > - * @instance: host adapter to clean up
> > - *
> > - * Free the generic interface resources from this adapter.
> > - *
> > - * Locks: none
> > - */
> > -
> > -static int generic_NCR5380_release_resources(struct Scsi_Host *instance)
> > +static void generic_NCR5380_release_resources(struct Scsi_Host
> > *instance) {
> > + scsi_remove_host(instance);
> > if (instance->irq != NO_IRQ)
> > free_irq(instance->irq, instance);
> > NCR5380_exit(instance);
> > @@ -364,7 +304,7 @@ static int generic_NCR5380_release_resources(struct
> > Scsi_Host *instance) release_mem_region(instance->base,
> > hostdata->iomem_size);
> > }
> > #endif
> > - return 0;
> > + scsi_host_put(instance);
>
> The sequence of operations here should be the same as the error path
> above.
I put scsi_host_put() call at the end because the release_mem_region code (in
the MMIO case) dereferences the hostdata pointer. I don't think it's safe to
do after scsi_host_put().
[...]
> > +static int pnp_registered;
> > +#endif /* !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP) */
> > +
> > +static int __init generic_NCR5380_init(void)
> > +{
> > + int ret = 0;
> > +
> > +#if !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP)
> > + ret = pnp_register_driver(&generic_NCR5380_pnp_driver);
> > + if (!ret)
> > + pnp_registered = 1;
> > #endif
> > + ret = isa_register_driver(&generic_NCR5380_isa_driver, MAX_CARDS);
> > + if (!ret)
> > + isa_registered = 1;
> > +
> > +#if !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP)
> > + if (pnp_registered)
> > + ret = 0;
> > +#endif
> > + if (isa_registered)
> > + ret = 0;
> > +
> > + return ret;
> > +}
> > +
> > +static void __exit generic_NCR5380_exit(void)
> > +{
> > +#if !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP)
> > + if (pnp_registered)
> > + pnp_unregister_driver(&generic_NCR5380_pnp_driver);
> > +#endif
> > + if (isa_registered)
> > + isa_unregister_driver(&generic_NCR5380_isa_driver);
> > +}
> > +
>
> I find that hard to follow. This should be equivalent and simpler:
>
> static int __init generic_NCR5380_init(void)
> {
> isa_ret = isa_register_driver(&generic_NCR5380_isa_driver, MAX_CARDS);
> #if !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP)
> pnp_ret = pnp_register_driver(&generic_NCR5380_pnp_driver);
> return pnp_ret ? isa_ret : 0;
> #endif
> return isa_ret;
> }
>
> static void __exit generic_NCR5380_exit(void)
> {
> if (!isa_ret)
> isa_unregister_driver(&generic_NCR5380_isa_driver);
> #if !defined(SCSI_G_NCR5380_MEM) && defined(CONFIG_PNP)
> if (!pnp_ret)
> pnp_unregister_driver(&generic_NCR5380_pnp_driver);
> #endif
> }
Doesn't make it any better, IMHO. Good that it's shorter but not cleaner -
especially this:
> return pnp_ret ? isa_ret : 0;
Also looking at the _exit function, meaning of isa_ret and pnp_ret is not
obvious there.
Maybe we should have a module_isa_pnp_driver() macro.
--
Ondrej Zary
next prev parent reply other threads:[~2016-09-27 12:55 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-24 18:59 [PATCH 0/3] g_NCR5380: Modernization Ondrej Zary
2016-09-24 18:59 ` [PATCH 1/3] g_NCR5380: Remove deprecated __setup Ondrej Zary
2016-09-25 23:45 ` Christoph Hellwig
2016-09-26 0:46 ` Finn Thain
2016-09-24 18:59 ` [PATCH 2/3] g_NCR5380: Reduce overrides[] from array to struct Ondrej Zary
2016-09-25 23:46 ` Christoph Hellwig
2016-09-26 3:56 ` Finn Thain
2016-09-24 18:59 ` [PATCH 3/3] g_NCR5380: Stop using scsi_module.c Ondrej Zary
2016-09-25 23:50 ` Christoph Hellwig
2016-09-26 3:29 ` Finn Thain
2016-09-27 12:55 ` Ondrej Zary [this message]
2016-09-28 0:06 ` Finn Thain
2016-09-25 19:39 ` [PATCH 4/3] g_NCR5380: Merge g_NCR5380 and g_NCR5380_mmio Ondrej Zary
2016-09-25 23:55 ` Christoph Hellwig
2016-09-26 0:37 ` Finn Thain
2016-09-26 0:49 ` Christoph Hellwig
2016-09-28 0:49 ` Finn Thain
-- strict thread matches above, loose matches on Subject: below --
2016-09-27 19:00 [PATCH v2 0/3] g_NCR5380: Modernization Ondrej Zary
2016-09-27 19:00 ` [PATCH 3/3] g_NCR5380: Stop using scsi_module.c Ondrej Zary
2016-09-28 22:57 ` Finn Thain
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=201609271455.17930.linux@rainbow-software.org \
--to=linux@rainbow-software.org \
--cc=fthain@telegraphics.com.au \
--cc=hch@infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).