I''m experiencing freezes with a -CURRENT of one and a half week ago with the COMMELL LE-565 board. To let the onboard CF card readed attached as wd0 to work I''ve to set 0x0ffc flags on wd* kernel device, plus to effectively use the four re nic I''ve to disable (comment out) pcibios kernel device. Does anyone else have tried a 4.4 system on these boards? Or have some hints on where i should look at? Thanks -- Massimo You definitely intend to start living sometime soon.
On Wed, 12 Nov 2008, Massimo Lusetti wrote:> I''m experiencing freezes with a -CURRENT of one and a half week ago > with the COMMELL LE-565 board. > > To let the onboard CF card readed attached as wd0 to work I''ve to set > 0x0ffc flags on wd* kernel device, plus to effectively use the four re > nic I''ve to disable (comment out) pcibios kernel device. > > Does anyone else have tried a 4.4 system on these boards? > Or have some hints on where i should look at?I had an LE565 until recently, but it stopped working with a puff of smoke :( It never had any trouble with CF, but it did have interrupt routing problems (long before 4.4) that went away with pcibios being turned off. You might want to try disabling apm instead and seeing if ACPI solves them - it didn''t work for me the last time I checked, but there have been many improvements since then. -d
On Wed, 12 Nov 2008 23:42:02 +1100 (EST) Damien Miller <djm at mindrot.org> wrote:> I had an LE565 until recently, but it stopped working with a puff of > smoke :(Really sorry for that> It never had any trouble with CF, but it did have interrupt routing > problems (long before 4.4) that went away with pcibios being turned > off. You might want to try disabling apm instead and seeing if ACPI > solves them - it didn''t work for me the last time I checked, but > there have been many improvements since then. > > -dWell with pcibios turned off the interrupts routes correctly but the CR reader still hangs with DMA issues. Did you set any particular value into the BIOS? I must say I''ve tried some without any luck. Thanks for any hints. -- Massimo Save energy: be apathetic.
On 2008/11/18 12:15, Massimo Lusetti wrote:> On Wed, 12 Nov 2008 23:42:02 +1100 (EST) > Damien Miller <djm at mindrot.org> wrote: > > > I had an LE565 until recently, but it stopped working with a puff of > > smoke :( > > Really sorry for that > > > It never had any trouble with CF, but it did have interrupt routing > > problems (long before 4.4) that went away with pcibios being turned > > off. You might want to try disabling apm instead and seeing if ACPI > > solves them - it didn''t work for me the last time I checked, but > > there have been many improvements since then. > > > > -d > > Well with pcibios turned off the interrupts routes correctly but the CR > reader still hangs with DMA issues. > > Did you set any particular value into the BIOS? I must say I''ve tried > some without any luck.The Commell probably doesn''t have DMA lines wired to the CF socket, but your card probably does support DMA. The OS can''t detect this other than by lack of response from the card after trying DMA. This problem is compounded by many CF cards which don''t correctly honour the RESET command sent when the OS tries to revert to PIO mode when it detects this situation. Try disabling DMA: boot -c, "change wd", and set the flags to 0x0ff0 - see wd(4) for more details. You can make it permanent with config(8) or by editing kernel config. It mightn''t be a bad idea for the flashboot config files to use these flags anyway..
On Tue, 18 Nov 2008 11:26:32 +0000 Stuart Henderson <stu at spacehopper.org> wrote:> The Commell probably doesn''t have DMA lines wired to the CF socket, > but your card probably does support DMA. The OS can''t detect this > other than by lack of response from the card after trying DMA. This > problem is compounded by many CF cards which don''t correctly honour > the RESET command sent when the OS tries to revert to PIO mode when > it detects this situation.That could be but i cannot understand why Damien didn''t experience the same problem with the same board...> Try disabling DMA: boot -c, "change wd", and set the flags to 0x0ff0 > - see wd(4) for more details.For what i know flashboot kernel didn''t support boot -c but...> You can make it permanent with config(8) or by editing kernel config.... I''ve slightly modified a flashboot kernel (even GENERIC-RD) to totally disable DMA (0x0ffc) and with that the kernel/OS boot regularly. I simply noted this issues coming out with >= 4.3 and not with my previous version which was based on 4.1> It mightn''t be a bad idea for the flashboot config files to use these > flags anyway..Yep... definitely since the flash isn''t accessed at all during normal operation. Regards -- Massimo Change your thoughts and you change your world.
On Tue, 18 Nov 2008 11:26:32 +0000 Stuart Henderson <stu at spacehopper.org> wrote:> Try disabling DMA: boot -c, "change wd", and set the flags to 0x0ff0 > - see wd(4) for more details.BTW this is the error with default wd flags: pciide0:0:1: bus-master DMA error: missing interrupt, status=0x41 wd1c: device timeout reading fsbn 0 (wd1 bn 0; cn 0 tn 0 sn 0), retrying wd1(pciide0:0:1): timeout type: ata c_bcount: 512 c_skip: 0 pciide0:0:1: bus-master DMA error: missing interrupt, status=0x41 wd1: transfer error, downgrading to Ultra-DMA mode 1 wd0(pciide0:0:0): using PIO mode 4 wd1(pciide0:0:1): using PIO mode 4, Ultra-DMA mode 1 wd1c: device timeout reading fsbn 0 (wd1 bn 0; cn 0 tn 0 sn 0), retrying Thanks for any hints. -- Massimo You are deeply attached to your friends and acquaintances.
On 2008/11/18 15:22, Massimo Lusetti wrote:> For what i know flashboot kernel didn''t support boot -c but...ah, I''m not sure about that, haven''t tried.> I simply noted this issues coming out with >= 4.3 and not with my > previous version which was based on 4.1DMA was not supported for most flash devices (except the majority of 1GB+ sandisk and one other brand I know of whose name escapes me) until 4.3. (specifically any device with 1-sector I/O).> BTW this is the error with default wd flags: > > pciide0:0:1: bus-master DMA error: missing interrupt, status=0x41 > wd1c: device timeout reading fsbn 0 (wd1 bn 0; cn 0 tn 0 sn 0), retrying > wd1(pciide0:0:1): timeout > type: ata > c_bcount: 512 > c_skip: 0 > pciide0:0:1: bus-master DMA error: missing interrupt, status=0x41 > wd1: transfer error, downgrading to Ultra-DMA mode 1 > wd0(pciide0:0:0): using PIO mode 4 > wd1(pciide0:0:1): using PIO mode 4, Ultra-DMA mode 1 > wd1c: device timeout reading fsbn 0 (wd1 bn 0; cn 0 tn 0 sn 0), retryingindeed, this is your device ignoring the reset sent when the OS downgrades. If you try one of the Sandisk cards which supports, it will almost certainly fall back correctly.
On Tue, 18 Nov 2008 14:48:45 +0000 Stuart Henderson <stu at spacehopper.org> wrote:> DMA was not supported for most flash devices (except the majority > of 1GB+ sandisk and one other brand I know of whose name escapes me) > until 4.3. (specifically any device with 1-sector I/O).That all make sense. Great.> indeed, this is your device ignoring the reset sent when the OS > downgrades. If you try one of the Sandisk cards which supports, > it will almost certainly fall back correctly.As i said that make sense, thanks for taking time to answer, your comments were really appreciated, thanks. Regards -- Massimo Good night to spend with family, but avoid arguments with your mate''s new lover.