From: Jiri Slaby <jirislaby@gmail.com>
To: Manu Abraham <manu@kromtek.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: PCI driver
Date: Wed, 14 Sep 2005 18:16:38 +0200 [thread overview]
Message-ID: <43284CE6.3080302@gmail.com> (raw)
In-Reply-To: <43281C27.1060305@kromtek.com>
Manu Abraham napsal(a):
> Jiri Slaby wrote:
>
>> Manu Abraham napsal(a):
>>
>>> Jiri Slaby wrote:
>>>
>>>> Manu Abraham napsal(a):
>>>>
>>>>> Jiri Slaby wrote:
>>>>>
>>>>>> Manu Abraham napsal(a):
>>>>>>
>>>>>>> Now that i have been trying to implement the driver using the
>>>>>>> new PCI API, i feel a bit lost at the different changes gone
>>>>>>> into the PCI API. So if someone could give me a brief idea how a
>>>>>>> minimal PCI probe routine should consist of, that would be quite
>>>>>>> helpful.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Maybe, you want to read http://lwn.net/Kernel/LDD3/, chapter 12,
>>>>>> pages 311+.
>>>>>>
>>>>>
>>>>> I have been updating myself from LDD2 to LDD3. What i was
>>>>> wondering was in what order should i be calling the functions.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> You won't call anything, kernel does. You only register driver.
>>>> struct pcitbl {venids, devids}
>>>>
>>>> struct driver ... = {probe=a, remove=b, tbl=pcitbl};
>>>> a() { if device from pcitbl is in the system (or has been added
>>>> before some little time) this function is called}
>>>> b() {if the device was removed from system: modules and hotplug:
>>>> never called; modules: so if modules unload; if both: if the device
>>>> was removed on the fly, or module unload}
>>>>
>>>> module_init() { register_driver(driver)}
>>>> module_exit() { unregister_driver(driver); }
>>>>
>>>
>>> I was wondering whether pci_enable_device() should come first or
>>> pci_dev_put() in the probe routine.
>>
>>
>> pci_dev_put? No, it counts down reference count, so you would loose
>> the structure. You do NOT do pci_dev_put anymore with pci probing
>> (but some very very specific cases).
>
>
>
> Oh, i thought after a pci_dev_get() one does a pci_dev_put()
> I wrote something like this, i was now confused with what to do with
> pci_get_drvdata() and pci_set_drvdata()
> I am not very sure whether i am doing it right in the first place, since
> my mapped memory seems to be wrong, but first i have to clear up about
> pci_get/set_drvdata().
>
>
> Manu
>
> static int mantis_pci_probe(*struct* pci_dev *pdev, const *struct*
> pci_device_id *mantis_pci_table)
> {
> *struct* mantis_pci *mantis;
> u8 revision, latency;
>
> pdev = pci_get_device(PCI_VENDOR_ID_MANTIS,
> PCI_DEVICE_ID_MANTIS_R11, NULL);
you do NOT do this at all, because you have pdev already (the param of
the probe function)
> *if* (pdev) {
> dprintk(verbose, MANTIS_DEBUG, 1, "Found a mantis chip");
> *if* ((mantis = (*struct* mantis_pci *) kmalloc(*sizeof*
> (*struct* mantis_pci), GFP_KERNEL)) == NULL) {
> dprintk(verbose, MANTIS_ERROR, 1, "Out of memory");
> *return* -ENOMEM;
> }
> *if* (pci_enable_device(pdev) < 0) {
you do only this
> dprintk(verbose, MANTIS_ERROR, 1, "Could not enable device");
> kfree(mantis);
> *return* -ENODEV;
gotos and return on one place
> }
> mantis->pdev = pdev;
you don't need this, need you? you put mantis with pci_setdrvdata and
whenever you want it, you get it from it
> mantis->mantis_addr = pci_resource_start(pdev, 0);
> *if* (!request_mem_region(pci_resource_start(pdev, 0),
> pci_resource_len(pdev, 0), DRIVER_NAME)) {
> kfree(mantis);
> *return* -EBUSY;
> }
> pci_read_config_byte(mantis->pdev, PCI_CLASS_REVISION,
> &revision);
> pci_read_config_byte(mantis->pdev, PCI_LATENCY_TIMER, &latency);
> mantis->mantis_mmio = ioremap(mantis->mantis_addr, 0x1000);
> mmwrite(0, MANTIS_INT_STAT); //* Clear interrupts *//
> *if* (request_irq(mantis->pdev->irq, (void *) mantis_pci_irq,
> SA_SHIRQ |
> SA_INTERRUPT, DRIVER_NAME, (void *) mantis) <
> 0) {
> dprintk(verbose, MANTIS_ERROR, 1, "Mantis IRQ registration
> failed");
> release_mem_region(pci_resource_start(mantis->pdev, 0),
> pci_resource_len(mantis->pdev, 0));
> pci_disable_device(pdev);
> kfree(mantis);
> }
> pci_dev_put(pdev);
don't do that
>
>
> }
>
>
> *return* 0;
> }
>
>
next prev parent reply other threads:[~2005-09-14 16:16 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-09-14 9:34 PCI driver Manu Abraham
2005-09-14 10:03 ` Jiri Slaby
2005-09-14 10:02 ` Manu Abraham
2005-09-14 10:29 ` Jiri Slaby
2005-09-14 11:53 ` Manu Abraham
2005-09-14 12:22 ` Jiri Slaby
2005-09-14 12:28 ` Manu Abraham
2005-09-14 12:48 ` Manu Abraham
2005-09-14 16:16 ` Jiri Slaby [this message]
2005-09-14 17:09 ` Manu Abraham
2005-09-14 19:00 ` Jiri Slaby
2005-09-14 19:00 ` Manu Abraham
2005-09-14 19:16 ` Jiri Slaby
2005-09-14 19:20 ` Manu Abraham
2005-09-14 22:27 ` Manu Abraham
2005-09-15 6:43 ` Rolf Eike Beer
2005-09-15 7:45 ` Manu Abraham
2005-09-15 8:18 ` Rolf Eike Beer
2005-09-15 8:51 ` Manu Abraham
2005-09-15 9:48 ` Rolf Eike Beer
2005-09-15 14:38 ` Manu Abraham
2005-09-15 14:57 ` Rolf Eike Beer
2005-09-15 16:59 ` Manu Abraham
2005-09-15 18:29 ` Manu Abraham
2005-09-15 10:29 ` Ralph Metzler
2005-09-15 10:35 ` Manu Abraham
2005-09-15 11:42 ` Manu Abraham
2005-09-15 12:08 ` Antonino A. Daplas
2005-09-15 12:24 ` Rolf Eike Beer
2005-09-15 12:32 ` Manu Abraham
2005-09-15 12:08 ` Rolf Eike Beer
2005-10-10 12:02 ` Rolf Eike Beer
2005-10-10 12:48 ` Manu Abraham
2005-10-10 13:25 ` Rolf Eike Beer
2005-10-10 13:16 ` Manu Abraham
[not found] ` <3888a5cd0510100719r3fddc368oa01e07e2c42b71e@mail.gmail.com>
2005-10-10 15:00 ` Manu Abraham
[not found] ` <3888a5cd0510100846p7f2ff70cid69a1136b9256ab6@mail.gmail.com>
2005-10-10 17:08 ` Manu Abraham
[not found] ` <4af2d03a0510101101n54ab0b1cvae177c3c992bf9a9@mail.gmail.com>
2005-10-10 18:28 ` Manu Abraham
[not found] ` <3888a5cd0510100725k579809a9o374930df9988bfa3@mail.gmail.com>
2005-10-10 15:02 ` Manu Abraham
[not found] ` <4af2d03a0510100528y236a1246tfc56c08a78f072d5@mail.gmail.com>
2005-10-10 12:51 ` Manu Abraham
-- strict thread matches above, loose matches on Subject: below --
2010-08-28 12:17 pci driver Srinivas Mankan
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=43284CE6.3080302@gmail.com \
--to=jirislaby@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=manu@kromtek.com \
/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