Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758249Ab0LTXA7 (ORCPT ); Mon, 20 Dec 2010 18:00:59 -0500 Received: from mail-bw0-f45.google.com ([209.85.214.45]:48986 "EHLO mail-bw0-f45.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758146Ab0LTXA5 convert rfc822-to-8bit (ORCPT ); Mon, 20 Dec 2010 18:00:57 -0500 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=RK+M1nA8cl+pHF4bxrV1Fc/0FBBvjIlSrwlzIHSD409LG1atj5mhaQFv+mpffWDqRP F0ahM9DdQyENTH+PRGIjg4kuT9UTZGKBbNLWjq2hHCnLU9ebRIqhXN6xUPtR/1PEnWP1 4AJkCXH444GekhA58XfSfWxIRnEh3LSiMMZRU= MIME-Version: 1.0 In-Reply-To: <4D0FD747.9020700@bluewatersys.com> References: <1289147348-31969-1-git-send-email-alchark@gmail.com> <20101107165745.GB1759@n2100.arm.linux.org.uk> <20101107171726.GF1759@n2100.arm.linux.org.uk> <20101108171930.GA1471@alchark-u3s.lan> <20101111212322.GA15533@alchark-u3s.lan> <20101111234957.GA28735@n2100.arm.linux.org.uk> <20101219174017.GA31832@alchark-u3s> <20101220191549.GH28157@n2100.arm.linux.org.uk> <20101220195423.GA14509@alchark-u3s> <4D0FC1AE.2050708@bluewatersys.com> <4D0FD747.9020700@bluewatersys.com> Date: Tue, 21 Dec 2010 02:00:54 +0300 Message-ID: Subject: Re: [PATCH 1/6 v9] ARM: Add basic architecture support for VIA/WonderMedia 85xx SoC's From: Alexey Charkov To: Ryan Mallon Cc: Russell King - ARM Linux , linux-arm-kernel@lists.infradead.org, vt8500-wm8505-linux-kernel@googlegroups.com, Eric Miao , =?UTF-8?Q?Uwe_Kleine=2DK=C3=B6nig?= , Albin Tonnerre , linux-kernel@vger.kernel.org Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 9108 Lines: 224 2010/12/21 Ryan Mallon : > On 12/21/2010 10:48 AM, Alexey Charkov wrote: >> 2010/12/20 Ryan Mallon : >>> On 12/21/2010 08:54 AM, Alexey Charkov wrote: >>>> This adds support for the family of Systems-on-Chip produced initially >>>> by VIA and now its subsidiary WonderMedia that have recently become >>>> widespread in lower-end Chinese ARM-based tablets and netbooks. >>>> >>>> Support is included for both VT8500 and WM8505, selectable by a >>>> configuration switch at kernel build time. >>>> >>>> Included are basic machine initialization files, register and >>>> interrupt definitions, support for the on-chip interrupt controller, >>>> high-precision OS timer, GPIO lines, necessary macros for early debug, >>>> pulse-width-modulated outputs control, as well as platform device >>>> configurations for the specific drivers implemented elsewhere. >>>> >>>> Signed-off-by: Alexey Charkov >>> >>> Hi Alexey, >>> >>> Quick review below. > >> >>>> +void __init wmt_set_resources(void) >>>> +{ >>>> +     resources_lcdc[0].start = wmt_current_regs->lcdc; >>>> +     resources_lcdc[0].end = wmt_current_regs->lcdc + SZ_1K - 1; >>>> +     resources_lcdc[1].start = wmt_current_irqs->lcdc; >>>> +     resources_lcdc[1].end = wmt_current_irqs->lcdc; >>> >>> Ah, this makes more sense. But why have all the indirection? The >>> wmt_regmaps table could just be replaced with #defines and then have >>> separate device files for the VT8500 and the WM8505. This would also >>> make clearer which variants have which peripherals. >>> >> >> This was the way I implemented it originally. However, Arnd made quite >> a valid suggestion to allow runtime selection of the chip variant, >> thus registers and interrupts need to be held in an indexed data type >> instead of just compile-time macros. In addition, there is now some >> overall movement towards unification of binary kernel images for >> different ARM variants (as far as I can see), so this would be >> required in any case. >> >> Furthermore, as with many unbranded Chinese products, it's somewhat >> difficult to reliably determine the exact chip version used in your >> netbook without disassembling it. Reading a hardware register for >> identification is easier :) > > Okay, that makes sense. I still think there must be a better way than > having a massive indirect table with all the values. Why not detect the > variant in the core code and then have something like: > > int init_devices(void) > { >        int board_type = detect_board_type(); > >        switch (board_type) { >        case BOARD_TYPE_VT8500: >                return vt8500_init_devices(); > >        case BOARD_TYPE_WM8505: >                return wm8500_init_devices(); >        } > >        pr_err("Unknown board type\n"); >        BUG(); /* panic()? */ >        while (1) >                ; > } > > Then you can have the peripheral setup for each of the variants in their > own files and use #defines. It may get tricky in a couple of places if > you need to be able to access some value directly which is different on > each of the variants, but that can be handled on a case by case basis. > The majority of the numbers will be passed into drivers via the resource > structs. > This is more or less what I'm doing right now - except for the separation between different files. I tried to avoid duplication of similar things here. Is the indirect table really a big issue? I'm a bit reluctant to copy about the whole devices.c for each chip variant, which would be otherwise required. Further, it would add more complexity to the timer, irq, gpio, i8042 and probably some other places. >>>> + >>>> +void __init vt8500_reserve_mem(void) >>>> +{ >>>> +#ifdef CONFIG_FB_VT8500 >>>> +     panels[current_panel_idx].bpp = 16; /* Always use RGB565 */ >>>> +     preallocate_fb(&panels[current_panel_idx], SZ_4M); >>>> +     vt8500_device_lcdc.dev.platform_data = &panels[current_panel_idx]; >>>> +#endif >>>> +} >>> >>> Not sure if this should exist in the platform code or the framebuffer >>> driver. In the latter case it would automatically be CONFIG_FB_VT8500 >>> and the platform_data can still be set in the platform code. Is there a >>> reason for this not to be in the framebuffer driver? >>> >> >> I can't reserve memory via memblock from the driver, and usual runtime >> allocation functions can't handle it (need alignment to 4 megabytes in >> 8500, framebuffer sizes exceed 4 megabytes in 8505). > > Can you use one of the initcalls from a driver to to the memblock > reserve? I don't know much about how memblock works. There are also the > various large page allocators in the works, but I don't think anything > has hit mainline yet. > No, drivers are initialized too late, even if compiled statically. Memblock can only be reserved through the designated machine callback, as far as I have understood. >>>> +void __init vt8500_init_irq(void) >>>> +{ >>>> +     unsigned int i; >>>> + >>>> +     ic_regbase = ioremap(wmt_current_regs->ic0, SZ_64K); >>>> + >>>> +     if (ic_regbase) { >>>> +             /* Enable rotating priority for IRQ */ >>>> +             writel((1 << 6), ic_regbase + 0x20); >>>> +             writel(0, ic_regbase + 0x24); >>>> + >>>> +             for (i = 0; i < wmt_current_irqs->nr_irqs; i++) { >>>> +                     /* Disable all interrupts and route them to IRQ */ >>>> +                     writeb(0x00, ic_regbase + VT8500_IC_DCTR + i); >>>> + >>>> +                     set_irq_chip(i, &vt8500_irq_chip); >>>> +                     set_irq_handler(i, handle_level_irq); >>>> +                     set_irq_flags(i, IRQF_VALID); >>>> +             } >>>> +     } else { >>>> +             printk(KERN_ERR "Unable to remap the Interrupt Controller " >>>> +                             "registers, not enabling IRQs!\n"); >>> >>> printk strings should be on a single line (can be > 80 columns) to make >>> grepping easier. You could also use the pr_ macros with pr_fmt set. >>> >> >> Well, checkpatch.pl complained about that in the first place, so I >> split the line. Should I merge them back in all instances? > > Yes. I think checkpatch has been changed to warn about spitting printk > strings across lines now. > Ok, will then merge strings. >>>> + >>>> +static inline void pwm_busy_wait(void __iomem *reg, u8 bitmask) >>>> +{ >>>> +     int loops = 1000; >>>> +     while ((readb(reg) & bitmask) && --loops) >>>> +             cpu_relax(); >>> >>> Ugh. If you are going to busy wait, can't you delay for a known amount >>> of time? Even better, can this be replaced with wait_event or some >>> equivalent? >>> >> >> The delay should be on the order of several bus cycles, where udelay >> actually busy-waits, too. wait_event would be longer than that to set >> up, and there is no associated interrupt. > > I meant if the hardware has some specific maximum wait time then you > could just delay that long. If there is no interrupt then wait_event and > friends probably aren't going to work. > > Maybe convert this to a timed loop (i.e. 1 second timeout) using > jiffies. That way you are never dependent on cpu speed. You should > probably also emit a warning if the timeout is reached and the device > still claims to be busy. > Agreed, will look into that. There is no definite maximum wait time, though (at least none that I know about). >>>> +} >>>> + >>>> +int pwm_config(struct pwm_device *pwm, int duty_ns, int period_ns) >>>> +{ >>>> +     unsigned long long c; >>>> +     unsigned long period_cycles, prescale, pv, dc; >>>> + >>>> +     if (pwm == NULL || period_ns == 0 || duty_ns > period_ns) >>>> +             return -EINVAL; >>>> + >>>> +     c = 25000000/2; /* wild guess --- need to implement clocks */ >>>> +     c = c * period_ns; >>>> +     do_div(c, 1000000000); >>>> +     period_cycles = c; >>> >>> This looks like it could be reworked to remove the do_div call. >>> >> >> I just followed PXA implementation in this regard. Are there any >> specific suggestions? Note that c should not be a complie-time >> constant eventually, as this is the guessed PWM base frequency (should >> be read from the hardware, but the code for clocks is not yet in). > > I didn't have a particular solution in mind, but often by changing the > units used and rearranging the math a bit you can often avoid having to > do horrible multiplies and divides. > > For now at least you could avoid the do_div by assigning period_cycles > directly. > It depends on period_ns, which is passed in as an argument from whatever uses PWM, so I'm not sure it can be assigned directly. Thanks, Alexey -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/