Re: [Qemu-devel] [PATCH v2 0/4] hw/ppc/spapr: Add "Processor Register Hypervisor Resource Access" H-calls
On 12.02.2016 01:31, David Gibson wrote: > On Thu, Feb 11, 2016 at 01:47:17PM +0100, Thomas Huth wrote: >> While we were recently debugging a problem with the H_SET_DABR >> call [1], I noticed that some hypercalls from the chapter 14.5.4.3 >> ("Processor Register Hypervisor Resource Access") from the LoPAPR >> spec [2] are still missing in QEMU. >> So here's are some patches that implement these hypercalls. Linux >> apparently does not depend on these hypercalls yet (otherwise somebody >> would have noticed this earlier), but the hypercalls are rather simple, >> so I think the implementations are quite straight-forward and easy to >> read. >> >> v2: >> - Don't use set_spr() and set cpu->env.spr[] directly instead >> - Completely reworked the H_PAGE_INIT patch to map the source >> and target pages for higher speed, and to be able to flush now >> the caches if requested. > > I've merged 1-3 into ppc-for-2.6. My only concern with those is > whether we need to be setting some extra flags in the ibm,hypertas > property now that they are implemented. Good point, ... looking at "Table 176. Hypervisor Call Function Table" in LoPAPR, it seems like there are indeed additional function sets that should be signaled with this property. But the current list seems to be in a bad shape, anyway, e.g. "hcall-debug" seems to be missing for the H_LOGICAL_CI_LOAD/STORE function. So I think we best fix this up with a separate patch later (I can submit one once I sorted out the H_PAGE_INIT patch). Thomas signature.asc Description: OpenPGP digital signature
[Qemu-devel] [PATCH v2 0/4] hw/ppc/spapr: Add "Processor Register Hypervisor Resource Access" H-calls
While we were recently debugging a problem with the H_SET_DABR call [1], I noticed that some hypercalls from the chapter 14.5.4.3 ("Processor Register Hypervisor Resource Access") from the LoPAPR spec [2] are still missing in QEMU. So here's are some patches that implement these hypercalls. Linux apparently does not depend on these hypercalls yet (otherwise somebody would have noticed this earlier), but the hypercalls are rather simple, so I think the implementations are quite straight-forward and easy to read. v2: - Don't use set_spr() and set cpu->env.spr[] directly instead - Completely reworked the H_PAGE_INIT patch to map the source and target pages for higher speed, and to be able to flush now the caches if requested. [1] http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=760a7364f27d974d [2] https://members.openpowerfoundation.org/document/dl/469 Thomas Huth (4): hw/ppc/spapr: Add h_set_sprg0 hypercall hw/ppc/spapr: Implement h_set_dabr hw/ppc/spapr: Implement the h_set_xdabr hypercall hw/ppc/spapr: Implement the h_page_init hypercall hw/ppc/spapr_hcall.c | 132 --- target-ppc/kvm_ppc.h | 24 +- 2 files changed, 147 insertions(+), 9 deletions(-) -- 1.8.3.1
Re: [Qemu-devel] [PATCH v2 0/4] hw/ppc/spapr: Add "Processor Register Hypervisor Resource Access" H-calls
On Thu, Feb 11, 2016 at 01:47:17PM +0100, Thomas Huth wrote: > While we were recently debugging a problem with the H_SET_DABR > call [1], I noticed that some hypercalls from the chapter 14.5.4.3 > ("Processor Register Hypervisor Resource Access") from the LoPAPR > spec [2] are still missing in QEMU. > So here's are some patches that implement these hypercalls. Linux > apparently does not depend on these hypercalls yet (otherwise somebody > would have noticed this earlier), but the hypercalls are rather simple, > so I think the implementations are quite straight-forward and easy to > read. > > v2: > - Don't use set_spr() and set cpu->env.spr[] directly instead > - Completely reworked the H_PAGE_INIT patch to map the source > and target pages for higher speed, and to be able to flush now > the caches if requested. I've merged 1-3 into ppc-for-2.6. My only concern with those is whether we need to be setting some extra flags in the ibm,hypertas property now that they are implemented. I still have some comments on 4/4. -- David Gibson| I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson signature.asc Description: PGP signature