Spectre and Cortex-A8/A9

Spectre and Cortex-A8/A9

[Adam Ford]

There is a lot of information pushed around the web about the Spectre
bug.  I went to ARM's support site and there are some patches
available from February that haven't made their way into the mainline.

Sorry if this has been covered before, but I am curious to know if
there is a good place or method to determine whether or not the
Cortex-A8 and A9 are still vulnerable and which versions or 4.4, 4.9,
or 4.14 have any items fixed.

Thanks for your patience.  :-)

adam

Spectre and Cortex-A8/A9

[Adam Ford]

There is a lot of information pushed around the web about the Spectre
bug.  I went to ARM's support site and there are some patches
available from February that haven't made their way into the mainline.

Sorry if this has been covered before, but I am curious to know if
there is a good place or method to determine whether or not the
Cortex-A8 and A9 are still vulnerable and which versions or 4.4, 4.9,
or 4.14 have any items fixed.

Thanks for your patience.  :-)

adam

Re: Spectre and Cortex-A8/A9

[Tony Lindgren]

* Adam Ford <aford173@gmail.com> [180509 18:23]:
> There is a lot of information pushed around the web about the Spectre
> bug.  I went to ARM's support site and there are some patches
> available from February that haven't made their way into the mainline.
> 
> Sorry if this has been covered before, but I am curious to know if
> there is a good place or method to determine whether or not the
> Cortex-A8 and A9 are still vulnerable and which versions or 4.4, 4.9,
> or 4.14 have any items fixed.

Well while we're waiting to figure out what all needs to be fixed,
here's what I'd do meanwhile:

1. Make sure the bootloader sets IBE bit

2. Make sure you have CONFIG_ARM_ERRATA_430973=y

3. Apply Marc Zyngier's patches from thread "[PATCH v4 0/6] 32bit ARM
   branch predictor hardening", but note that these are still being
   discussed

4. Update kernel when patches that have been validated against
   test cases get merged

Regards,

Tony

Spectre and Cortex-A8/A9

[Tony Lindgren]

* Adam Ford <aford173@gmail.com> [180509 18:23]:
> There is a lot of information pushed around the web about the Spectre
> bug.  I went to ARM's support site and there are some patches
> available from February that haven't made their way into the mainline.
> 
> Sorry if this has been covered before, but I am curious to know if
> there is a good place or method to determine whether or not the
> Cortex-A8 and A9 are still vulnerable and which versions or 4.4, 4.9,
> or 4.14 have any items fixed.

Well while we're waiting to figure out what all needs to be fixed,
here's what I'd do meanwhile:

1. Make sure the bootloader sets IBE bit

2. Make sure you have CONFIG_ARM_ERRATA_430973=y

3. Apply Marc Zyngier's patches from thread "[PATCH v4 0/6] 32bit ARM
   branch predictor hardening", but note that these are still being
   discussed

4. Update kernel when patches that have been validated against
   test cases get merged

Regards,

Tony