From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sergei Shtylyov Subject: Re: [PATCH] siimage: fix ->set_pio_mode method to select PIO data transfer Date: Sat, 14 Jul 2007 21:35:50 +0400 Message-ID: <46990976.5020308@ru.mvista.com> References: <200707132311.58022.bzolnier@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from h155.mvista.com ([63.81.120.155]:50651 "EHLO imap.sh.mvista.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1757630AbXGNRdr (ORCPT ); Sat, 14 Jul 2007 13:33:47 -0400 In-Reply-To: <200707132311.58022.bzolnier@gmail.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Bartlomiej Zolnierkiewicz Cc: linux-ide@vger.kernel.org Bartlomiej Zolnierkiewicz wrote: > * Remember to select PIO data transfer (with IORDY monitored) in sil_tune_pio() > (->set_pio_mode method) so the controller is always programmed correctly for > PIO transfers (this is important if DMA is not going to be used). > * Don't set DMA/UDMA timings for PIO modes in siimage_tune_chipset(). > * Bump driver version. > Signed-off-by: Bartlomiej Zolnierkiewicz Acked-by: Sergei Shtylyov > Index: b/drivers/ide/pci/siimage.c > =================================================================== > --- a/drivers/ide/pci/siimage.c > +++ b/drivers/ide/pci/siimage.c > @@ -1,5 +1,5 @@ > /* > - * linux/drivers/ide/pci/siimage.c Version 1.15 Jun 29 2007 > + * linux/drivers/ide/pci/siimage.c Version 1.16 Jul 13 2007 > * > * Copyright (C) 2001-2002 Andre Hedrick > * Copyright (C) 2003 Red Hat > @@ -185,7 +185,12 @@ static void sil_tune_pio(ide_drive_t *dr > u16 speedp = 0; > unsigned long addr = siimage_seldev(drive, 0x04); > unsigned long tfaddr = siimage_selreg(hwif, 0x02); > + unsigned long base = (unsigned long)hwif->hwif_data; > u8 tf_pio = pio; > + u8 addr_mask = hwif->channel ? (hwif->mmio ? 0xF4 : 0x84) > + : (hwif->mmio ? 0xB4 : 0x80); > + u8 mode = 0; > + u8 unit = drive->select.b.unit; > > /* trim *taskfile* PIO to the slowest of the master/slave */ > if (pair->present) { > @@ -207,6 +212,11 @@ static void sil_tune_pio(ide_drive_t *dr > hwif->OUTW(hwif->INW(tfaddr-2)|0x200, tfaddr-2); > else > hwif->OUTW(hwif->INW(tfaddr-2)&~0x200, tfaddr-2); > + > + mode = hwif->INB(base + addr_mask); > + mode &= ~(unit ? 0x30 : 0x03); > + mode |= (unit ? 0x10 : 0x01); > + hwif->OUTB(mode, base + addr_mask); > } else { > pci_write_config_word(hwif->pci_dev, addr, speedp); > pci_write_config_word(hwif->pci_dev, tfaddr, speedt); > @@ -216,6 +226,11 @@ static void sil_tune_pio(ide_drive_t *dr > if (pio > 2) > speedp |= 0x200; > pci_write_config_word(hwif->pci_dev, tfaddr-2, speedp); > + > + pci_read_config_byte(hwif->pci_dev, addr_mask, &mode); > + mode &= ~(unit ? 0x30 : 0x03); > + mode |= (unit ? 0x10 : 0x01); > + pci_write_config_byte(hwif->pci_dev, addr_mask, mode); > } > } First, there's a possibility of disabling DMA mode if an invalid PIO mode is passed -- ide_config_drive_speed() won't call dma_off_quietly() in this case. Well, due to the way IDE core works now, one has to re-tune DMA after setting any PIO mode anyway, so maybe this simple code would indeed be enough... Second, although being correct (only for the drives supporting IORDY, that is), this code is asymmetric to what's done for the taskfile IORDY. A side note: pio/mmio code patchs certalinly could use some reorg here, to extract the common code from them... at least that's what was on my mind when I intended to work on this driver. :-) MBR, Sergei