| Message ID | 20201023071022.24916-1-kraxel@redhat.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | arm/virt: add usb support | expand |
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
---
tests/data/acpi/microvm/DSDT.usb | Bin 414 -> 426 bytes
tests/data/acpi/virt/DSDT.usb | Bin 0 -> 5257 bytes
2 files changed, 0 insertions(+), 0 deletions(-)
diff --git a/tests/data/acpi/microvm/DSDT.usb b/tests/data/acpi/microvm/DSDT.usb
index 14423381ea235ed42f6f0f7d98e793c271a4e7c1..bceef778d2633793d5192846e02fb4719de503a0 100644
GIT binary patch
delta 57
zcmbQoyo#C2CD<io6(a)!)6<Du(o8Pm8`aGjrL7}8oIN?>Jv?1_953(+aKwjtx-f9W
NJ3BivPR?aC1OT2T4c`C&
delta 45
zcmZ3*Jdc^nCD<io9wP$-Q_(~&X(kV`jq2u%{Q40d&Ym3c9-b~dju&_ZCRZ{V0s!~<
B3f}+#
diff --git a/tests/data/acpi/virt/DSDT.usb b/tests/data/acpi/virt/DSDT.usb
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..a5a46e044bb788da9028a572f55bc28036ce38bf 100644
GIT binary patch
literal 5257
zcmZvg%WoT16o>EFlh__VGIpHz!zq*s3RHeHP5LDE*oj@@#CX!CQYyI;0#UN5M1@o;
zq!6J(yQ7g<2PqQ&1`;b4NZob8nhpN|JIuLrXOiDh&qx{1nYrhi`;Cv|<J<I`JI{+s
z#r|w;du@NIaj)k!>71pM(#QJM{eIoLKkyray~A0rTrMXE?M^dW_j(-)R#b9ur_)q>
zhu6IAZgj2RZI{cHN_Z_f@cMp!)|q;iQW}_2u2RoOm1Uiy=n3VnuJ3y5hWkmYdxyAH
z%Jp~ZPMVUvx@8b~Cub0`)O46xx9|3&DQ#}`w<&e`?6#+wXE%3LEWCLc5fyWHj?lM#
z0})eah)RfP=vh=HgcFk(HZZElsELfY#25ue4H=h3hAlCYz_^TzWsxx=F&tnlBjbw5
z7?l_qU|d1QipWSvj2tjlkg+N<k`luO#ws$dij0)RC;;OsFe>4PgO`}&NQ@#dt|3D#
zQ~n`|N_`R-*O4LDX-1mIG%#);L$1@T#Fz!fO=QS*nv)pwz_^7BxlYF<#v(92LWW$Y
zuEbaZ#u_r@I?YRrw}4SchFqrwiE$nn4P?l5IxaCT0K)@DRj$*b#CR7NO=QS*Iw3JG
z0;7ctxlSh~#s|PyM}}OdQxfA6FgB1O*XgvxC<CL747pBcBt{h&9c0LLIx8_Q1LHO_
z<T{;`7*~MtF*4*jotGG^z}Q5FT&D{X;~Fr!z^KV}x+pPj0AmXoa-EhW#w}oMhm5UX
zdtN7;M{lcFu9sJ-`S7ScGEq%;F|7^SHyT@~ZHd3r&zGF}p7K;jeWsRFnf{t|^tF1Z
zKBZTsAwlEO>+g(je&rk#(sp_{R2zP$tjw`R#}!q`=<`mDf@&1RDX6JJ*3JxvmeM&i
zE{t>*4a}KqK^_f_LY+zcU@Ub>W^5Sg%xEaJK>Cg}G{~e-XGV942bs|f;!0;mL*pis
zh6b4w>P(tZa2GqsjAmd5nbA;cnL^7TlS1?yVV)yFX5=|yGNZ{;h@PX&lbTYRPc)QT
zEcB!hJrm3`!8{Y(lS1@NGS4LQq?QalDMZf{^Q3m#JiXwNq_`)A=;<&|hj~&9hn^Io
zXPSAYnP-}NQiz@z=9yuh)bgPxh3J`Oo>}IZ<(?FxXO4O1m?yP}=t&`Zjxo<M<~hba
zDMU~DIb&)(sA89SQcH=R6ryLIdFGjCo_kV=o(1MvV4l>1q9=vuInF%Cnddn7q!2yn
zhaz|unJ2ZZ=t&`ZPB702<~hMVDMZgn<~hkcsl`Q43ej_lc}_9UDeg%jdeV<e@SJ9z
z)Doj7h3GlMJZG5a4ELlEJ!hHcEc2um8a*jQ&pGBf$2{k_Cxz%a&phXuC$-$@Ng;YJ
zFwX_%xxhUsM9)R$xyU@JMMqBx(X+%nOU$#xJt;Jvdxsrczm?|aoBmSv%Il|J#64P;
z$T{g?DdO$mE1kY~-#Aws9r%6U3Q=Llz)-YKfT;5V%A7kz-PW@T8(fCc>jlVqM_`$A
zoE0{@oJH#f$hsh~%(;+NbO&Ceu1@HGsB+h6XC6m8;XM8^t77!;oVKCfSwGOHD?{By
zu(`wGfBI(`ZNp1+#mZ0*U^Z%6XdBX1OTBm>a)QeT`kLN?=xnFYKT%2lPOqnu-cEm0
z71vXPZQrwB>?^OiTRCngj=so`JMLchuiC$*`L&;SAJ>1o`TFx0`;JWms`Q$q*V-c;
zXpTOpY1C7_b4q%D?7lpDXnjjpD#7T=@#M>+Xq-9i>Wh!`V152%|CgLOuWwR0*)M)L
zX!>3I_?(%|tCc_g_Lslu{xtPZcjkAmdz?D@Qk~@EmZf*J(ABm3EuU4RUg**_S98j5
zJfGk_pr=z>J9{+xh`w*kx>+M}|B_yr+h{chro_#%TtzpYv@DmU3nAasdE9!>q#m^T
z)R)pbnoK$N-gjvodM&@Txl~*jwC?Vt9_ir0v3vCNfusAqXT9Bi!#UVHyne6kb%L8^
j>aVh{S^gu3pkHMRed&~13)k8%h7+FA6#OuHW}@nU`oteS
literal 0
HcmV?d00001
On Fri, 23 Oct 2020 at 08:10, Gerd Hoffmann <kraxel@redhat.com> wrote: > > Bring new microvm goodies to arm virt too. Wire up > -machine usb=on, add sysbus-xhci in case it is enabled. So my question here is the usual one -- why can't we just use a PCI USB controller ? thanks -- PMM
On Fri, Oct 23, 2020 at 09:10:14AM +0200, Gerd Hoffmann wrote: > Bring new microvm goodies to arm virt too. Wire up > -machine usb=on, add sysbus-xhci in case it is enabled. For ACPI bits: Reviewed-by: Michael S. Tsirkin <mst@redhat.com> > Gerd Hoffmann (8): > tests/acpi: allow updates for expected data files > tests/acpi: add empty tests/data/acpi/virt/DSDT.usb file > arm/virt: add support for -machine usb=on > arm/virt: add device tree node for xhci > arm/virt: add acpi dsdt entry for xhci > tests/acpi: add usb testcase for virt > tests/acpi: update expected data files > tests/acpi: disallow updates for expected data files > > include/hw/arm/virt.h | 1 + > hw/arm/virt-acpi-build.c | 6 ++++++ > hw/arm/virt.c | 36 +++++++++++++++++++++++++++++++ > hw/usb/hcd-xhci-sysbus.c | 2 ++ > tests/qtest/bios-tables-test.c | 18 ++++++++++++++++ > hw/arm/Kconfig | 1 + > tests/data/acpi/microvm/DSDT.usb | Bin 414 -> 426 bytes > tests/data/acpi/virt/DSDT.usb | Bin 0 -> 5257 bytes > 8 files changed, 64 insertions(+) > create mode 100644 tests/data/acpi/virt/DSDT.usb > > -- > 2.27.0 >
On Fri, Oct 23, 2020 at 12:36:05PM +0100, Peter Maydell wrote: > On Fri, 23 Oct 2020 at 08:10, Gerd Hoffmann <kraxel@redhat.com> wrote: > > > > Bring new microvm goodies to arm virt too. Wire up > > -machine usb=on, add sysbus-xhci in case it is enabled. > > So my question here is the usual one -- why can't we > just use a PCI USB controller ? Well, pci seems to come with some extra resource needs (see -M pc vs. -M q35 memory footprint differences discussed some months ago). Thats why microvm started without pci support, and even now with pcie support added it is optional (and off by default). I'm wondering whenever it makes sense for arm/virt to make pcie optional too. Adding an OnOffAuto pcie switch is easy. Some places which assume pci is present need fixing (-cdrom for example blindly uses virtio-blk-pci). Looks doable without too much effort and it would effectively bring a microvm-ish machine type to the arm/virt world. So in case pcie is switchable using sysbus-xhci would bring usb support without requiring pcie support for that. With pcie being present unconditionally there isn't much of a difference between sysbus-xhci and qemu-xhci (the pci variant of the device). The only problem with machine,usb=on adding a qemu-xhci device automatically would be that we would have to pick some pci slot where we can place the device ... take care, Gerd
On Mon, 26 Oct 2020 at 07:01, Gerd Hoffmann <kraxel@redhat.com> wrote: > On Fri, Oct 23, 2020 at 12:36:05PM +0100, Peter Maydell wrote: > > On Fri, 23 Oct 2020 at 08:10, Gerd Hoffmann <kraxel@redhat.com> wrote: > > > > > > Bring new microvm goodies to arm virt too. Wire up > > > -machine usb=on, add sysbus-xhci in case it is enabled. > > > > So my question here is the usual one -- why can't we > > just use a PCI USB controller ? > > Well, pci seems to come with some extra resource needs (see -M pc vs. > -M q35 memory footprint differences discussed some months ago). Thats > why microvm started without pci support, and even now with pcie support > added it is optional (and off by default). I think PCI is too useful to discard. You can run anything you want (practically) via PCI. If we make it optional then we're going to give ourselves the task of reimplementing memory mapped versions of all the functionality it gives us, all of which is extra code and all of which adds to the amount of stuff on the guest-to-QEMU security boundary. I think to be persuaded that no-PCI was a good idea I'd want to at least see solid figures based on doing this for Arm and on having put a lot of effort into slimming down the PCI handling code/overhead in the guest and still not being able to get near an MMIO setup. That is, try the "improve the guest" approach first. thanks -- PMM
Hi, > > Well, pci seems to come with some extra resource needs (see -M pc vs. > > -M q35 memory footprint differences discussed some months ago). Thats > > why microvm started without pci support, and even now with pcie support > > added it is optional (and off by default). > > I think PCI is too useful to discard. Discard is out of question of course. I'd only add the option to disable it if not needed. > You can run anything you want (practically) via PCI. If we make it > optional then we're going to give ourselves the task of reimplementing > memory mapped versions of all the functionality it gives us, No. Well, at least it would not be *my* plan to reach feature parity. I'm just trying to make available what we have. The mmio versions of usb host adapters are needed anyway to emulate some SoCs. Describing them in device tree / ACPI is standardized so they just work without additional changes on the guest side. So this is really just adding the device to the machine, adding a device tree node, adding a acpi dsdt entry (roughly a dozen lines of code each). Having virtio(-mmio) and usb is enough to cover alot of use cases. Especially on arm where virtio-gpu is the only display device option. > all of which is extra code and all of which adds to the > amount of stuff on the guest-to-QEMU security boundary. usb is off by default so it doesn't add anything unless you explicitly ask for it. Oh, and pci adds some non-trivial stuff to the guest-to-QEMU security boundary too ... > I think to be persuaded that no-PCI was a good idea I'd > want to at least see solid figures based on doing this > for Arm and on having put a lot of effort into slimming > down the PCI handling code/overhead in the guest and still > not being able to get near an MMIO setup. That is, try the > "improve the guest" approach first. Ok, fair enough. take care, Gerd