| Message ID | 20240731055022.696051-1-adityag@linux.ibm.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Power11 support for QEMU [PSeries] | expand |
Hi, Any comments on this ? Thanks, Aditya Gupta On 31/07/24 11:20, Aditya Gupta wrote: > Overview > ============ > > Split "Power11 support for QEMU" into 2 patch series: pseries & powernv. > > This patch series is for pseries support for Power11. > > As Power11 core is same as Power10, hence much of the code has been reused from > Power10. > > Power11 was added in Linux in: > commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode") > > Git Tree for Testing > ==================== > > QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries > > Has been tested with following cases: > * '-M pseries' / '-M pseries -cpu Power11' > * '-smp' option tested > * with compat mode: 'max-cpu-compat=power10' and 'max-cpu-compat=power9' > * with/without device 'virtio-scsi-pci' > * with/without -kernel and -drive with qcow_file > > Linux with Power11 support: https://github.com/torvalds/linux, since v6.9-rc1 > > Changelog > ========= > v6 RESEND: > + added my initials instead of PMM in patch #1 description > > v6: > + reorganised patches such that Power11 introduction is at end, and > cleanups and fixes is done before > + patch #1: renamed macros from POWER_* to PCC_* > + patch #2: rename 'logical_pvr' to 'spapr_logical_pvr' to better convey > the context > > v5: > + split patch series into pseries+powernv > + patch #1: apply harsh's patch to reduce duplication > + patch #2: simplified, by removing duplication > + patch #3: update docs, according to harsh's suggestion > + patch #4: no functional change, #define used for P9 & P10 pcr_supported > + patch #5: no change > > v4: > + patch #5: fix memory leak in pnv_chip_power10_quad_realize > - no change in other patches > > v3: > + patch #1: version power11 as power11_v2.0 > + patch #2: split target hw/pseries code into patch #2 > + patch #3,#4: fix regression due to Power10 and Power11 having same PCR > + patch #5: create pnv_chip_power11_dt_populate and split pnv_chip_power10_common_realize as per review > + patch #6-#11: no change > - remove commit to make Power11 as default > > v2: > + split powernv patch into homer,lpc,occ,psi,sbe > + reduce code duplication by reusing power10 code > + make power11 as default > + rebase on qemu upstream/master > + add more information in commit descriptions > + update docs > + update skiboot.lid > > > Aditya Gupta (4): > target/ppc: Introduce 'PowerPCCPUClass::spapr_logical_pvr' > target/ppc: Fix regression due to Power10 and Power11 having same PCR > target/ppc: Add Power11 DD2.0 processor > ppc/pseries: Add Power11 cpu type > > Harsh Prateek Bora (1): > target/ppc: Reduce code duplication across Power9/10 init code > > docs/system/ppc/pseries.rst | 17 +++- > hw/ppc/spapr_cpu_core.c | 1 + > target/ppc/compat.c | 11 +++ > target/ppc/cpu-models.c | 3 + > target/ppc/cpu-models.h | 3 + > target/ppc/cpu.h | 3 + > target/ppc/cpu_init.c | 188 +++++++++++++++--------------------- > target/ppc/cpu_init.h | 92 ++++++++++++++++++ > 8 files changed, 205 insertions(+), 113 deletions(-) > create mode 100644 target/ppc/cpu_init.h >
Hi Aditya, On 2024/07/31 11:20 AM, Aditya Gupta wrote: > Overview > ============ > > Split "Power11 support for QEMU" into 2 patch series: pseries & powernv. > > This patch series is for pseries support for Power11. > > As Power11 core is same as Power10, hence much of the code has been reused from > Power10. > > Power11 was added in Linux in: > commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode") > > Git Tree for Testing > ==================== > > QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries > > Has been tested with following cases: > * '-M pseries' / '-M pseries -cpu Power11' I tried the below command with mailine QEMU and this patch series applied. I could boot the guest with -cpu Power11 option but when I check inside the guest with `lscpu`, I can still see the Model name being reported as "Power 10" while the PVR value looks fine corresponding to Power11. Did you see "Power 11" while you were testing this? $ qemu-system-ppc64 -m 4G -smp 4 -nographic -drive file=/root/testing/debian-12-generic-ppc64el.qcow2,format=qcow2 -accel tcg -cpu Power11 root@localhost:~# lscpu Architecture: ppc64le Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Model name: POWER10 (architected), altivec supported Model: 18.0 (pvr 0082 1200) Thanks, Amit > * '-smp' option tested > * with compat mode: 'max-cpu-compat=power10' and 'max-cpu-compat=power9' > * with/without device 'virtio-scsi-pci' > * with/without -kernel and -drive with qcow_file > > Linux with Power11 support: https://github.com/torvalds/linux, since v6.9-rc1 > > Changelog > ========= > v6 RESEND: > + added my initials instead of PMM in patch #1 description > > v6: > + reorganised patches such that Power11 introduction is at end, and > cleanups and fixes is done before > + patch #1: renamed macros from POWER_* to PCC_* > + patch #2: rename 'logical_pvr' to 'spapr_logical_pvr' to better convey > the context > > v5: > + split patch series into pseries+powernv > + patch #1: apply harsh's patch to reduce duplication > + patch #2: simplified, by removing duplication > + patch #3: update docs, according to harsh's suggestion > + patch #4: no functional change, #define used for P9 & P10 pcr_supported > + patch #5: no change > > v4: > + patch #5: fix memory leak in pnv_chip_power10_quad_realize > - no change in other patches > > v3: > + patch #1: version power11 as power11_v2.0 > + patch #2: split target hw/pseries code into patch #2 > + patch #3,#4: fix regression due to Power10 and Power11 having same PCR > + patch #5: create pnv_chip_power11_dt_populate and split pnv_chip_power10_common_realize as per review > + patch #6-#11: no change > - remove commit to make Power11 as default > > v2: > + split powernv patch into homer,lpc,occ,psi,sbe > + reduce code duplication by reusing power10 code > + make power11 as default > + rebase on qemu upstream/master > + add more information in commit descriptions > + update docs > + update skiboot.lid > > > Aditya Gupta (4): > target/ppc: Introduce 'PowerPCCPUClass::spapr_logical_pvr' > target/ppc: Fix regression due to Power10 and Power11 having same PCR > target/ppc: Add Power11 DD2.0 processor > ppc/pseries: Add Power11 cpu type > > Harsh Prateek Bora (1): > target/ppc: Reduce code duplication across Power9/10 init code > > docs/system/ppc/pseries.rst | 17 +++- > hw/ppc/spapr_cpu_core.c | 1 + > target/ppc/compat.c | 11 +++ > target/ppc/cpu-models.c | 3 + > target/ppc/cpu-models.h | 3 + > target/ppc/cpu.h | 3 + > target/ppc/cpu_init.c | 188 +++++++++++++++--------------------- > target/ppc/cpu_init.h | 92 ++++++++++++++++++ > 8 files changed, 205 insertions(+), 113 deletions(-) > create mode 100644 target/ppc/cpu_init.h > > -- > 2.45.2 > >
Hi Amit, On 29/08/24 17:49, Amit Machhiwal wrote: > Hi Aditya, > > On 2024/07/31 11:20 AM, Aditya Gupta wrote: >> Overview >> ============ >> >> Split "Power11 support for QEMU" into 2 patch series: pseries & powernv. >> >> This patch series is for pseries support for Power11. >> >> As Power11 core is same as Power10, hence much of the code has been reused from >> Power10. >> >> Power11 was added in Linux in: >> commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode") >> >> Git Tree for Testing >> ==================== >> >> QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries >> >> Has been tested with following cases: >> * '-M pseries' / '-M pseries -cpu Power11' > I tried the below command with mailine QEMU and this patch series applied. I > could boot the guest with -cpu Power11 option but when I check inside the guest > with `lscpu`, I can still see the Model name being reported as "Power 10" while > the PVR value looks fine corresponding to Power11. > > Did you see "Power 11" while you were testing this? > > $ qemu-system-ppc64 -m 4G -smp 4 -nographic -drive file=/root/testing/debian-12-generic-ppc64el.qcow2,format=qcow2 -accel tcg -cpu Power11 > > root@localhost:~# lscpu > Architecture: ppc64le > Byte Order: Little Endian > CPU(s): 4 > On-line CPU(s) list: 0-3 > Model name: POWER10 (architected), altivec supported > Model: 18.0 (pvr 0082 1200) Yes. Above system's kernel is probably older, and doesn't know about Power11. Power11 support was added in upstream linux, with commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode"). Also, if you notice the model is being printed as 0x00821200, it is the Processor version register (PVR), the 0x82 in it is for Power11. If it had been Power10, it would be 0x80. You can try that with -cpu Power10. You will need to use a kernel newer than v6.9-rc1, which contains the commit adding Power11 support to the kernel. Thanks, Aditya Gupta > > Thanks, > Amit > >> * '-smp' option tested >> * with compat mode: 'max-cpu-compat=power10' and 'max-cpu-compat=power9' >> * with/without device 'virtio-scsi-pci' >> * with/without -kernel and -drive with qcow_file >> >> Linux with Power11 support: https://github.com/torvalds/linux, since v6.9-rc1 >> >> Changelog >> ========= >> v6 RESEND: >> + added my initials instead of PMM in patch #1 description >> >> v6: >> + reorganised patches such that Power11 introduction is at end, and >> cleanups and fixes is done before >> + patch #1: renamed macros from POWER_* to PCC_* >> + patch #2: rename 'logical_pvr' to 'spapr_logical_pvr' to better convey >> the context >> >> v5: >> + split patch series into pseries+powernv >> + patch #1: apply harsh's patch to reduce duplication >> + patch #2: simplified, by removing duplication >> + patch #3: update docs, according to harsh's suggestion >> + patch #4: no functional change, #define used for P9 & P10 pcr_supported >> + patch #5: no change >> >> v4: >> + patch #5: fix memory leak in pnv_chip_power10_quad_realize >> - no change in other patches >> >> v3: >> + patch #1: version power11 as power11_v2.0 >> + patch #2: split target hw/pseries code into patch #2 >> + patch #3,#4: fix regression due to Power10 and Power11 having same PCR >> + patch #5: create pnv_chip_power11_dt_populate and split pnv_chip_power10_common_realize as per review >> + patch #6-#11: no change >> - remove commit to make Power11 as default >> >> v2: >> + split powernv patch into homer,lpc,occ,psi,sbe >> + reduce code duplication by reusing power10 code >> + make power11 as default >> + rebase on qemu upstream/master >> + add more information in commit descriptions >> + update docs >> + update skiboot.lid >> >> >> Aditya Gupta (4): >> target/ppc: Introduce 'PowerPCCPUClass::spapr_logical_pvr' >> target/ppc: Fix regression due to Power10 and Power11 having same PCR >> target/ppc: Add Power11 DD2.0 processor >> ppc/pseries: Add Power11 cpu type >> >> Harsh Prateek Bora (1): >> target/ppc: Reduce code duplication across Power9/10 init code >> >> docs/system/ppc/pseries.rst | 17 +++- >> hw/ppc/spapr_cpu_core.c | 1 + >> target/ppc/compat.c | 11 +++ >> target/ppc/cpu-models.c | 3 + >> target/ppc/cpu-models.h | 3 + >> target/ppc/cpu.h | 3 + >> target/ppc/cpu_init.c | 188 +++++++++++++++--------------------- >> target/ppc/cpu_init.h | 92 ++++++++++++++++++ >> 8 files changed, 205 insertions(+), 113 deletions(-) >> create mode 100644 target/ppc/cpu_init.h >> >> -- >> 2.45.2 >> >>
On 2024/07/31 11:20 AM, Aditya Gupta wrote: > Overview > ============ > > Split "Power11 support for QEMU" into 2 patch series: pseries & powernv. > > This patch series is for pseries support for Power11. > > As Power11 core is same as Power10, hence much of the code has been reused from > Power10. > > Power11 was added in Linux in: > commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode") > > Git Tree for Testing > ==================== > > QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries > > Has been tested with following cases: > * '-M pseries' / '-M pseries -cpu Power11' > I could boot a Power11 pseries KVM guest with this patch series applied which does not work without it. Command Used: qemu-system-ppc64 -m 4G -smp 4 -nographic -drive file=/root/testing/debian-12-generic-ppc64el.qcow2,format=qcow2 -accel tcg -cpu Power11 Tested-by: Amit Machhiwal <amachhiw@linux.ibm.com> > * '-smp' option tested > * with compat mode: 'max-cpu-compat=power10' and 'max-cpu-compat=power9' > * with/without device 'virtio-scsi-pci' > * with/without -kernel and -drive with qcow_file > > Linux with Power11 support: https://github.com/torvalds/linux, since v6.9-rc1 > > Changelog > ========= > v6 RESEND: > + added my initials instead of PMM in patch #1 description > > v6: > + reorganised patches such that Power11 introduction is at end, and > cleanups and fixes is done before > + patch #1: renamed macros from POWER_* to PCC_* > + patch #2: rename 'logical_pvr' to 'spapr_logical_pvr' to better convey > the context > > v5: > + split patch series into pseries+powernv > + patch #1: apply harsh's patch to reduce duplication > + patch #2: simplified, by removing duplication > + patch #3: update docs, according to harsh's suggestion > + patch #4: no functional change, #define used for P9 & P10 pcr_supported > + patch #5: no change > > v4: > + patch #5: fix memory leak in pnv_chip_power10_quad_realize > - no change in other patches > > v3: > + patch #1: version power11 as power11_v2.0 > + patch #2: split target hw/pseries code into patch #2 > + patch #3,#4: fix regression due to Power10 and Power11 having same PCR > + patch #5: create pnv_chip_power11_dt_populate and split pnv_chip_power10_common_realize as per review > + patch #6-#11: no change > - remove commit to make Power11 as default > > v2: > + split powernv patch into homer,lpc,occ,psi,sbe > + reduce code duplication by reusing power10 code > + make power11 as default > + rebase on qemu upstream/master > + add more information in commit descriptions > + update docs > + update skiboot.lid > > > Aditya Gupta (4): > target/ppc: Introduce 'PowerPCCPUClass::spapr_logical_pvr' > target/ppc: Fix regression due to Power10 and Power11 having same PCR > target/ppc: Add Power11 DD2.0 processor > ppc/pseries: Add Power11 cpu type > > Harsh Prateek Bora (1): > target/ppc: Reduce code duplication across Power9/10 init code > > docs/system/ppc/pseries.rst | 17 +++- > hw/ppc/spapr_cpu_core.c | 1 + > target/ppc/compat.c | 11 +++ > target/ppc/cpu-models.c | 3 + > target/ppc/cpu-models.h | 3 + > target/ppc/cpu.h | 3 + > target/ppc/cpu_init.c | 188 +++++++++++++++--------------------- > target/ppc/cpu_init.h | 92 ++++++++++++++++++ > 8 files changed, 205 insertions(+), 113 deletions(-) > create mode 100644 target/ppc/cpu_init.h > > -- > 2.45.2 > >
Hi Amit, On 02/09/24 11:19, Amit Machhiwal wrote: > On 2024/07/31 11:20 AM, Aditya Gupta wrote: >> <...snip...> >> >> Git Tree for Testing >> ==================== >> >> QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries >> >> Has been tested with following cases: >> * '-M pseries' / '-M pseries -cpu Power11' >> > I could boot a Power11 pseries KVM guest with this patch series applied which > does not work without it. > > Command Used: > > qemu-system-ppc64 -m 4G -smp 4 -nographic -drive file=/root/testing/debian-12-generic-ppc64el.qcow2,format=qcow2 -accel tcg -cpu Power11 > > Tested-by: Amit Machhiwal <amachhiw@linux.ibm.com> Thanks for testing it, Amit ! - Aditya Gupta
Overview ============ Split "Power11 support for QEMU" into 2 patch series: pseries & powernv. This patch series is for pseries support for Power11. As Power11 core is same as Power10, hence much of the code has been reused from Power10. Power11 was added in Linux in: commit c2ed087ed35c ("powerpc: Add Power11 architected and raw mode") Git Tree for Testing ==================== QEMU: https://github.com/adi-g15-ibm/qemu/tree/p11-v6-pseries Has been tested with following cases: * '-M pseries' / '-M pseries -cpu Power11' * '-smp' option tested * with compat mode: 'max-cpu-compat=power10' and 'max-cpu-compat=power9' * with/without device 'virtio-scsi-pci' * with/without -kernel and -drive with qcow_file Linux with Power11 support: https://github.com/torvalds/linux, since v6.9-rc1 Changelog ========= v6 RESEND: + added my initials instead of PMM in patch #1 description v6: + reorganised patches such that Power11 introduction is at end, and cleanups and fixes is done before + patch #1: renamed macros from POWER_* to PCC_* + patch #2: rename 'logical_pvr' to 'spapr_logical_pvr' to better convey the context v5: + split patch series into pseries+powernv + patch #1: apply harsh's patch to reduce duplication + patch #2: simplified, by removing duplication + patch #3: update docs, according to harsh's suggestion + patch #4: no functional change, #define used for P9 & P10 pcr_supported + patch #5: no change v4: + patch #5: fix memory leak in pnv_chip_power10_quad_realize - no change in other patches v3: + patch #1: version power11 as power11_v2.0 + patch #2: split target hw/pseries code into patch #2 + patch #3,#4: fix regression due to Power10 and Power11 having same PCR + patch #5: create pnv_chip_power11_dt_populate and split pnv_chip_power10_common_realize as per review + patch #6-#11: no change - remove commit to make Power11 as default v2: + split powernv patch into homer,lpc,occ,psi,sbe + reduce code duplication by reusing power10 code + make power11 as default + rebase on qemu upstream/master + add more information in commit descriptions + update docs + update skiboot.lid Aditya Gupta (4): target/ppc: Introduce 'PowerPCCPUClass::spapr_logical_pvr' target/ppc: Fix regression due to Power10 and Power11 having same PCR target/ppc: Add Power11 DD2.0 processor ppc/pseries: Add Power11 cpu type Harsh Prateek Bora (1): target/ppc: Reduce code duplication across Power9/10 init code docs/system/ppc/pseries.rst | 17 +++- hw/ppc/spapr_cpu_core.c | 1 + target/ppc/compat.c | 11 +++ target/ppc/cpu-models.c | 3 + target/ppc/cpu-models.h | 3 + target/ppc/cpu.h | 3 + target/ppc/cpu_init.c | 188 +++++++++++++++--------------------- target/ppc/cpu_init.h | 92 ++++++++++++++++++ 8 files changed, 205 insertions(+), 113 deletions(-) create mode 100644 target/ppc/cpu_init.h