From: Hartvig Ekner <hartvig@ekner.info>
To: Pete Popov <ppopov@mvista.com>
Cc: Linux MIPS mailing list <linux-mips@linux-mips.org>
Subject: Re: Patch to disable PCI coherency on AU1500 platforms
Date: Tue, 01 Apr 2003 22:07:23 +0200 [thread overview]
Message-ID: <3E89F17B.5EFA2E37@ekner.info> (raw)
In-Reply-To: 1049221843.26884.256.camel@zeus.mvista.com
Sure, no problem waiting - I don't have the problem any more :-) However, if you don't have this
fix in your kernel, you should see massive PCI transfer errors (as I did), which can explain why your "cp -a" fails.
OTOH, I can't explain why you don't see the same problem on the older 2.4.18 kernel you mentioned
was working just fine.
HPT371 also works here (== it boots without crashing midstream during boot) after updating to the
latest CVS with the new IDE code. I will run some stress tests to see if the residual errors (even
with PCI_CFG[NC] set) also occur on this controller, or only on the Promise.
/Hartvig
Pete Popov wrote:
> I have a Pb1500 but like I said, I won't be able to get to it until end
> of the month. I would really like to run some stress tests before
> applying such core patches. Can you wait until I get back?
>
> Pete
>
> On Tue, 2003-04-01 at 04:30, Hartvig Ekner wrote:
> > The patch below sets the NC bit in the PCI_CFG register to disable HW coherency when
> > running non-coherent. Until now, this bit was cleared which means corruption when using PCI
> > DMA masters, even if the kernel was correctly compiled with CONFIG_NONCOHERENT_IO.
> >
> > Pb1500 specific notes: I don't have a PB1500, so I cannot test if it works there. Note: I also
> > removed what I think was an extraneous write to the PCI_CMEM register, so if somebody
> > could test this on a PB1500 it would be great.
> >
> > /Hartvig
> >
> >
> >
> > ______________________________________________________________________
> >
> > Index: db1x00/setup.c
> > ===================================================================
> > RCS file: /home/cvs/linux/arch/mips/au1000/db1x00/Attic/setup.c,v
> > retrieving revision 1.1.2.4
> > diff -u -r1.1.2.4 setup.c
> > --- db1x00/setup.c 21 Mar 2003 19:00:46 -0000 1.1.2.4
> > +++ db1x00/setup.c 1 Apr 2003 12:14:54 -0000
> > @@ -78,9 +78,8 @@
> > void __init au1x00_setup(void)
> > {
> > char *argptr;
> > - u32 pin_func, static_cfg0;
> > - u32 sys_freqctrl, sys_clksrc;
> > - u32 prid = read_c0_prid();
> > + u32 pin_func;
> > +// u32 prid = read_c0_prid();
> >
> > argptr = prom_getcmdline();
> >
> > @@ -187,6 +186,19 @@
> >
> > #ifdef CONFIG_BLK_DEV_IDE
> > ide_ops = &std_ide_ops;
> > +#endif
> > +
> > +#ifdef CONFIG_PCI
> > + /* Although YAMON has setup the PCI controller, some things
> > + may need to change. Eventually, all the PCI initialization
> > + should be done here (as in eg. ../pb1500/setup.c)
> > + */
> > +
> > +#ifdef CONFIG_NONCOHERENT_IO
> > + /* Must disable PCI coherency if running non-coherent */
> > +
> > + au_writel(au_readl(Au1500_PCI_CFG) | (1<<16), Au1500_PCI_CFG);
> > +#endif
> > #endif
> >
> > #if 0
> > Index: pb1500/setup.c
> > ===================================================================
> > RCS file: /home/cvs/linux/arch/mips/au1000/pb1500/setup.c,v
> > retrieving revision 1.1.2.12
> > diff -u -r1.1.2.12 setup.c
> > --- pb1500/setup.c 21 Mar 2003 19:00:47 -0000 1.1.2.12
> > +++ pb1500/setup.c 1 Apr 2003 12:14:54 -0000
> > @@ -35,6 +35,7 @@
> > #include <linux/console.h>
> > #include <linux/mc146818rtc.h>
> > #include <linux/delay.h>
> > +#include <linux/proc_fs.h>
> >
> > #include <asm/cpu.h>
> > #include <asm/bootinfo.h>
> > @@ -90,6 +91,7 @@
> > char *argptr;
> > u32 pin_func, static_cfg0;
> > u32 sys_freqctrl, sys_clksrc;
> > + u32 pcicfg;
> >
> > argptr = prom_getcmdline();
> >
> > @@ -232,15 +234,25 @@
> >
> > #ifdef CONFIG_PCI
> > // Setup PCI bus controller
> > - au_writel(0, Au1500_PCI_CMEM);
> > - au_writel(0x00003fff, Au1500_CFG_BASE);
> > +
> > + au_writel(0x00003fff, Au1500_PCI_CMEM);
> > +
> > #if defined(__MIPSEB__)
> > - au_writel(0xf | (2<<6) | (1<<4), Au1500_PCI_CFG);
> > + pcicfg = 0xf | (2<<6) | (1<<4);
> > #else
> > - au_writel(0xf, Au1500_PCI_CFG);
> > + pcicfg = 0xf;
> > #endif
> > +
> > +#ifdef CONFIG_NONCOHERENT_IO
> > + /* Must disable PCI coherency if running non-coherent */
> > +
> > + pcicfg |= (1<<16);
> > +#endif
> > +
> > + au_writel(pcicfg, Au1500_PCI_CFG);
> > +
> > au_writel(0xf0000000, Au1500_PCI_MWMASK_DEV);
> > - au_writel(0, Au1500_PCI_MWBASE_REV_CCL);
> > + au_writel(0, Au1500_PCI_MWBASE_REV_CCL);
> > au_writel(0x02a00356, Au1500_PCI_STATCMD);
> > au_writel(0x00003c04, Au1500_PCI_HDRTYPE);
> > au_writel(0x00000008, Au1500_PCI_MBAR);
next prev parent reply other threads:[~2003-04-01 20:00 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-04-01 12:30 Patch to disable PCI coherency on AU1500 platforms Hartvig Ekner
2003-04-01 18:30 ` Pete Popov
2003-04-01 20:07 ` Hartvig Ekner [this message]
2003-04-01 20:40 ` Pete Popov
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=3E89F17B.5EFA2E37@ekner.info \
--to=hartvig@ekner.info \
--cc=linux-mips@linux-mips.org \
--cc=ppopov@mvista.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