Hi Rick,
I have ported your BadRAM patch to the new kernel 2.6.14. There were
a few tiny formal corrections due to patch conflicts; besides that, I
did not change anything.
To linux-kernel: is there a reason why this patch was never added to
Linus' tree? It helped me save money more than once.
Max
On 11/9/05, Max Kellermann <[email protected]> wrote:
> Hi Rick,
>
> I have ported your BadRAM patch to the new kernel 2.6.14. There were
> a few tiny formal corrections due to patch conflicts; besides that, I
> did not change anything.
>
> To linux-kernel: is there a reason why this patch was never added to
> Linus' tree? It helped me save money more than once.
>
> Max
>
>
>
Hey,
Didn't happen to make a x86_64 one did you? Really good job, I love
this patch so much and I think I'll look into a 64bit variant when I'm
bored in french or something.
-DaMouse
DaMouse wrote:
> On 11/9/05, Max Kellermann <[email protected]> wrote:
>
>>Hi Rick,
>>
>>I have ported your BadRAM patch to the new kernel 2.6.14. There were
>>a few tiny formal corrections due to patch conflicts; besides that, I
>>did not change anything.
>>
>>To linux-kernel: is there a reason why this patch was never added to
>>Linus' tree? It helped me save money more than once.
Might be because memmap boot parameter (Documentation/kernel-parameters.txt) is sufficient for most
cases?
I used a machine with 1GB RAM and 3 holes of about 5MB each for almost 2 years.
> Didn't happen to make a x86_64 one did you? Really good job, I love
> this patch so much and I think I'll look into a 64bit variant when I'm
> bored in french or something.
If this get included, a 64bit version would be appreciated though.
Kalin.
--
|[ ~~~~~~~~~~~~~~~~~~~~~~ ]|
+-> http://ThinRope.net/ <-+
|[ ______________________ ]|
On Wed, 9 Nov 2005, Max Kellermann wrote:
> I have ported your BadRAM patch to the new kernel 2.6.14. There were
> a few tiny formal corrections due to patch conflicts; besides that, I
> did not change anything.
>
> To linux-kernel: is there a reason why this patch was never added to
> Linus' tree? It helped me save money more than once.
We would like to do something similar but would include bad RAM removal
while the operating system is running. This may be because single bit ECC
failures occur (which would cause the page to migrate away and mark the
page as bad) or a hard ECC failure (references to the page are dropped if
possible and the page is not dirty)