The Annoying Affair - Bulletin #1

The Annoying Affair - Bulletin #1

[Thomas Gleixner]

[-- Attachment #1: Type: text/plain, Size: 619 bytes --]

Folks,

I've merged the V2 of Boris' cleaned up variant of the TAA series. Thanks
Boris for handling that!

The master branch is based on v5.4-rc5 and the commits are considered
stable from now on. Any further tweaks need to go on top.

The stable branches linux-5.3.y, linux-4.19.y and linix-4.14.y have the
backports. These need some careful inspection and testing.

Dead kernels are left as an exercise for the members of the Kernel
Necrophilia Cult as always.

The changes are available as usual from:

  cvs.ou.linutronix.de:linux/speck/linux  

Git bundles for all relevant branches are attached.

Thanks,

	tglx

[-- Attachment #2: Type: application/octet-stream, Size: 23609 bytes --]

[-- Attachment #3: Type: application/octet-stream, Size: 23996 bytes --]

[-- Attachment #4: Type: application/octet-stream, Size: 24951 bytes --]

[-- Attachment #5: Type: application/octet-stream, Size: 24954 bytes --]

[MODERATED] Re: The Annoying Affair - Bulletin #1

[Pawan Gupta]

On Mon, Oct 28, 2019 at 09:31:40AM +0100, speck for Thomas Gleixner wrote:
> The stable branches linux-5.3.y, linux-4.19.y and linix-4.14.y have the
> backports. These need some careful inspection and testing.

Intel is testing these backports.

For commit "kvm/x86: Export MDS_NO=0 to guests when TSX is enabled"
backport to 4.4 has dependency on below upstream commits.

0cf9135b773bf32fba9dd8e6699c1b331ee4b749 KVM: x86: Emulate MSR_IA32_ARCH_CAPABILITIES on AMD hosts 
5b76a3cff011df2dcb6186c965a2e4d809a05ad4 KVM: VMX: Tell the nested hypervisor to skip L1D flush on vmentry

There is a minor fix for 4.14 backport.

---
diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index f668e10a2669..8a7eccf7ba7b 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -31,6 +31,8 @@
 #include <asm/intel-family.h>
 #include <asm/e820.h>
 
+#include "cpu.h"
+
 static void __init spectre_v1_select_mitigation(void);
 static void __init spectre_v2_select_mitigation(void);
 static void __init ssb_select_mitigation(void);

[MODERATED] Re: The Annoying Affair - Bulletin #1

[mark gross]

On Mon, Oct 28, 2019 at 09:31:40AM +0100, speck for Thomas Gleixner wrote:
> Folks,
> 
> I've merged the V2 of Boris' cleaned up variant of the TAA series. Thanks
> Boris for handling that!
> 
> The master branch is based on v5.4-rc5 and the commits are considered
> stable from now on. Any further tweaks need to go on top.
> 
> The stable branches linux-5.3.y, linux-4.19.y and linix-4.14.y have the
> backports. These need some careful inspection and testing.
Neelima, our testing lead, is getting her spec-list access enabled so I'm
sending this quick note on the testing so far:
5.3.y backport checked out.
4.19.y backport is showing some issues with disabling TSX.  We are
investigating and will report out more detail / patch update ASAP.

--mark

> 
> Dead kernels are left as an exercise for the members of the Kernel
> Necrophilia Cult as always.
> 
> The changes are available as usual from:
> 
>   cvs.ou.linutronix.de:linux/speck/linux  
> 
> Git bundles for all relevant branches are attached.
> 
> Thanks,
> 
> 	tglx

[MODERATED] Re: The Annoying Affair - Bulletin #1

[Pawan Gupta]

On Tue, Oct 29, 2019 at 09:14:27AM -0700, speck for mark gross wrote:
> On Mon, Oct 28, 2019 at 09:31:40AM +0100, speck for Thomas Gleixner wrote:
> > Folks,
> > 
> > I've merged the V2 of Boris' cleaned up variant of the TAA series. Thanks
> > Boris for handling that!
> > 
> > The master branch is based on v5.4-rc5 and the commits are considered
> > stable from now on. Any further tweaks need to go on top.
> > 
> > The stable branches linux-5.3.y, linux-4.19.y and linix-4.14.y have the
> > backports. These need some careful inspection and testing.
> Neelima, our testing lead, is getting her spec-list access enabled so I'm
> sending this quick note on the testing so far:
> 5.3.y backport checked out.
> 4.19.y backport is showing some issues with disabling TSX.  We are
> investigating and will report out more detail / patch update ASAP.

4.14 and 4.19 needs this change as well, 5.3 is fine.

---
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 97f79547782e..a05637f6d95b 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1497,6 +1497,7 @@ void __init identify_boot_cpu(void)
        enable_sep_cpu();
 #endif
        cpu_detect_tlb(&boot_cpu_data);
+       tsx_init();
 }
 
 void identify_secondary_cpu(struct cpuinfo_x86 *c)

[MODERATED] Re: The Annoying Affair - Bulletin #1

[Pawan Gupta]

[-- Attachment #1: Type: text/plain, Size: 1661 bytes --]

On Tue, Oct 29, 2019 at 01:00:38PM -0700, speck for Pawan Gupta wrote:
> On Tue, Oct 29, 2019 at 09:14:27AM -0700, speck for mark gross wrote:
> > On Mon, Oct 28, 2019 at 09:31:40AM +0100, speck for Thomas Gleixner wrote:
> > > Folks,
> > > 
> > > I've merged the V2 of Boris' cleaned up variant of the TAA series. Thanks
> > > Boris for handling that!
> > > 
> > > The master branch is based on v5.4-rc5 and the commits are considered
> > > stable from now on. Any further tweaks need to go on top.
> > > 
> > > The stable branches linux-5.3.y, linux-4.19.y and linix-4.14.y have the
> > > backports. These need some careful inspection and testing.
> > Neelima, our testing lead, is getting her spec-list access enabled so I'm
> > sending this quick note on the testing so far:
> > 5.3.y backport checked out.
> > 4.19.y backport is showing some issues with disabling TSX.  We are
> > investigating and will report out more detail / patch update ASAP.
> 
> 4.14 and 4.19 needs this change as well, 5.3 is fine.
> 
> ---
> diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
> index 97f79547782e..a05637f6d95b 100644
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -1497,6 +1497,7 @@ void __init identify_boot_cpu(void)
>         enable_sep_cpu();
>  #endif
>         cpu_detect_tlb(&boot_cpu_data);
> +       tsx_init();
>  }
>  
>  void identify_secondary_cpu(struct cpuinfo_x86 *c)

Attaching bundles with these minor changes included. All backports are
tested to be working fine. 4.4 is not ported yet, KVM patches have
multiple unmet dependencies.

Thanks,
Pawan

[-- Attachment #2: taa-4.14.bundle --]
[-- Type: application/octet-stream, Size: 25067 bytes --]

[-- Attachment #3: taa-4.19.bundle --]
[-- Type: application/octet-stream, Size: 25020 bytes --]

[-- Attachment #4: taa-4.9.bundle --]
[-- Type: application/octet-stream, Size: 24734 bytes --]

[-- Attachment #5: taa-5.3.bundle --]
[-- Type: application/octet-stream, Size: 23968 bytes --]