Re: [linux-yocto] [PATCH 2/2] intel-x86-64: Move 'CONFIG_NR_CPUS=256' to intel-x86-64.cfg
On 11/09/2018 10:06 PM, Bruce Ashfield wrote: On 2018-11-08 8:45 p.m., Hongzhi, Song wrote: On 11/09/2018 03:56 AM, Bruce Ashfield wrote: On 2018-11-07 9:05 p.m., Hongzhi.Song wrote: The maximum cpus are 64 on intel-x86-32. But intel-x86-64 support ranges from 1 to 8192. So we should move the config to intel-x86-64.cfg. ok. This makes more sense, but two questions: - why limit to 256 on x86-64 ? Hi Bruce, The config file which contains '256' was transplanted from the historical branch. I have consulted the committer, Yongxin Liu , who said: for most x86-64 products, 256 was an usual value. [add Yongxin in loop] - why not put 64 as the limit in intel-x86.cfg, since it now has no setting at all. The default value is 64 on intel-x86-32. So I doesn't explicitly set the config to 64 in intel-x86.cfg, but set 256 explicitly in intel-x86-64. I'd still rather have it explicitly set. We do sometimes rely on defaults, but it is often better to have the fragments be self-documenting and be independent of changing defaults in the kernels. By the way, intel-x86.cfg is shared by 32bit and 64bit. Indeed. That would be a problem in setting the value to 64 explicitly in that fragment. I'll merge this as-is, but I would rather have something that explicitly sets the value for 32 bit .. something for a future re-org of the fragments to take care of. Ok. After your merge, I will add 'CONFIG_NR_CPU=64' to intel-x86-32.cfg. Thanks. --Hongzhi Bruce --Hongzhi Bruce Signed-off-by: Hongzhi.Song --- bsp/intel-x86/intel-x86-64.cfg | 3 +++ bsp/intel-x86/intel-x86.cfg | 1 - 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/bsp/intel-x86/intel-x86-64.cfg b/bsp/intel-x86/intel-x86-64.cfg index 4e8a4d7..215c0f0 100644 --- a/bsp/intel-x86/intel-x86-64.cfg +++ b/bsp/intel-x86/intel-x86-64.cfg @@ -49,3 +49,6 @@ CONFIG_CRYPTO_DEV_QAT_DH895xCCVF=m # Intel Resource Director Technology support CONFIG_INTEL_RDT=y + +# Processor type and features +CONFIG_NR_CPUS=256 diff --git a/bsp/intel-x86/intel-x86.cfg b/bsp/intel-x86/intel-x86.cfg index 6919179..080bca1 100644 --- a/bsp/intel-x86/intel-x86.cfg +++ b/bsp/intel-x86/intel-x86.cfg @@ -17,7 +17,6 @@ CONFIG_MCORE2=y CONFIG_SMP=y CONFIG_SCHED_SMT=y -CONFIG_NR_CPUS=256 CONFIG_NUMA=y CONFIG_ACPI_NUMA=y -- ___ linux-yocto mailing list linux-yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/linux-yocto
Re: [linux-yocto] [PATCH 2/2] intel-x86-64: Move 'CONFIG_NR_CPUS=256' to intel-x86-64.cfg
On 2018-11-08 8:45 p.m., Hongzhi, Song wrote: On 11/09/2018 03:56 AM, Bruce Ashfield wrote: On 2018-11-07 9:05 p.m., Hongzhi.Song wrote: The maximum cpus are 64 on intel-x86-32. But intel-x86-64 support ranges from 1 to 8192. So we should move the config to intel-x86-64.cfg. ok. This makes more sense, but two questions: - why limit to 256 on x86-64 ? Hi Bruce, The config file which contains '256' was transplanted from the historical branch. I have consulted the committer, Yongxin Liu , who said: for most x86-64 products, 256 was an usual value. [add Yongxin in loop] - why not put 64 as the limit in intel-x86.cfg, since it now has no setting at all. The default value is 64 on intel-x86-32. So I doesn't explicitly set the config to 64 in intel-x86.cfg, but set 256 explicitly in intel-x86-64. I'd still rather have it explicitly set. We do sometimes rely on defaults, but it is often better to have the fragments be self-documenting and be independent of changing defaults in the kernels. By the way, intel-x86.cfg is shared by 32bit and 64bit. Indeed. That would be a problem in setting the value to 64 explicitly in that fragment. I'll merge this as-is, but I would rather have something that explicitly sets the value for 32 bit .. something for a future re-org of the fragments to take care of. Bruce --Hongzhi Bruce Signed-off-by: Hongzhi.Song --- bsp/intel-x86/intel-x86-64.cfg | 3 +++ bsp/intel-x86/intel-x86.cfg | 1 - 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/bsp/intel-x86/intel-x86-64.cfg b/bsp/intel-x86/intel-x86-64.cfg index 4e8a4d7..215c0f0 100644 --- a/bsp/intel-x86/intel-x86-64.cfg +++ b/bsp/intel-x86/intel-x86-64.cfg @@ -49,3 +49,6 @@ CONFIG_CRYPTO_DEV_QAT_DH895xCCVF=m # Intel Resource Director Technology support CONFIG_INTEL_RDT=y + +# Processor type and features +CONFIG_NR_CPUS=256 diff --git a/bsp/intel-x86/intel-x86.cfg b/bsp/intel-x86/intel-x86.cfg index 6919179..080bca1 100644 --- a/bsp/intel-x86/intel-x86.cfg +++ b/bsp/intel-x86/intel-x86.cfg @@ -17,7 +17,6 @@ CONFIG_MCORE2=y CONFIG_SMP=y CONFIG_SCHED_SMT=y -CONFIG_NR_CPUS=256 CONFIG_NUMA=y CONFIG_ACPI_NUMA=y -- ___ linux-yocto mailing list linux-yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/linux-yocto
Re: [linux-yocto] [PATCH 2/2] intel-x86-64: Move 'CONFIG_NR_CPUS=256' to intel-x86-64.cfg
On 11/09/2018 03:56 AM, Bruce Ashfield wrote: On 2018-11-07 9:05 p.m., Hongzhi.Song wrote: The maximum cpus are 64 on intel-x86-32. But intel-x86-64 support ranges from 1 to 8192. So we should move the config to intel-x86-64.cfg. ok. This makes more sense, but two questions: - why limit to 256 on x86-64 ? Hi Bruce, The config file which contains '256' was transplanted from the historical branch. I have consulted the committer, Yongxin Liu , who said: for most x86-64 products, 256 was an usual value. [add Yongxin in loop] - why not put 64 as the limit in intel-x86.cfg, since it now has no setting at all. The default value is 64 on intel-x86-32. So I doesn't explicitly set the config to 64 in intel-x86.cfg, but set 256 explicitly in intel-x86-64. By the way, intel-x86.cfg is shared by 32bit and 64bit. --Hongzhi Bruce Signed-off-by: Hongzhi.Song --- bsp/intel-x86/intel-x86-64.cfg | 3 +++ bsp/intel-x86/intel-x86.cfg | 1 - 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/bsp/intel-x86/intel-x86-64.cfg b/bsp/intel-x86/intel-x86-64.cfg index 4e8a4d7..215c0f0 100644 --- a/bsp/intel-x86/intel-x86-64.cfg +++ b/bsp/intel-x86/intel-x86-64.cfg @@ -49,3 +49,6 @@ CONFIG_CRYPTO_DEV_QAT_DH895xCCVF=m # Intel Resource Director Technology support CONFIG_INTEL_RDT=y + +# Processor type and features +CONFIG_NR_CPUS=256 diff --git a/bsp/intel-x86/intel-x86.cfg b/bsp/intel-x86/intel-x86.cfg index 6919179..080bca1 100644 --- a/bsp/intel-x86/intel-x86.cfg +++ b/bsp/intel-x86/intel-x86.cfg @@ -17,7 +17,6 @@ CONFIG_MCORE2=y CONFIG_SMP=y CONFIG_SCHED_SMT=y -CONFIG_NR_CPUS=256 CONFIG_NUMA=y CONFIG_ACPI_NUMA=y -- ___ linux-yocto mailing list linux-yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/linux-yocto
Re: [linux-yocto] [PATCH 2/2] intel-x86-64: Move 'CONFIG_NR_CPUS=256' to intel-x86-64.cfg
On 2018-11-07 9:05 p.m., Hongzhi.Song wrote: The maximum cpus are 64 on intel-x86-32. But intel-x86-64 support ranges from 1 to 8192. So we should move the config to intel-x86-64.cfg. ok. This makes more sense, but two questions: - why limit to 256 on x86-64 ? - why not put 64 as the limit in intel-x86.cfg, since it now has no setting at all. Bruce Signed-off-by: Hongzhi.Song --- bsp/intel-x86/intel-x86-64.cfg | 3 +++ bsp/intel-x86/intel-x86.cfg| 1 - 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/bsp/intel-x86/intel-x86-64.cfg b/bsp/intel-x86/intel-x86-64.cfg index 4e8a4d7..215c0f0 100644 --- a/bsp/intel-x86/intel-x86-64.cfg +++ b/bsp/intel-x86/intel-x86-64.cfg @@ -49,3 +49,6 @@ CONFIG_CRYPTO_DEV_QAT_DH895xCCVF=m # Intel Resource Director Technology support CONFIG_INTEL_RDT=y + +# Processor type and features +CONFIG_NR_CPUS=256 diff --git a/bsp/intel-x86/intel-x86.cfg b/bsp/intel-x86/intel-x86.cfg index 6919179..080bca1 100644 --- a/bsp/intel-x86/intel-x86.cfg +++ b/bsp/intel-x86/intel-x86.cfg @@ -17,7 +17,6 @@ CONFIG_MCORE2=y CONFIG_SMP=y CONFIG_SCHED_SMT=y -CONFIG_NR_CPUS=256 CONFIG_NUMA=y CONFIG_ACPI_NUMA=y -- ___ linux-yocto mailing list linux-yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/linux-yocto