Re: [PATCH 01/14] spapr: Simplify error handling in spapr_phb_realize()
On Thu, Aug 13, 2020 at 11:39:27PM +0200, Greg Kurz wrote:
> On Thu, 13 Aug 2020 16:57:04 -0300
> Daniel Henrique Barboza wrote:
>
> > Greg,
> >
> > This patch is breaking guest startup in ppc-for-5.2 for me. The process
> > gives an almost instant segfault. Here's what I'm doing:
> >
>
> Ok, this is because this patch should have been applied after
> the "spapr: Cleanups for XIVE" series:
>
> http://patchwork.ozlabs.org/project/qemu-devel/cover/159679991916.876294.8967140647442842745.st...@bahia.lan/
>
> specifically this patch:
>
> http://patchwork.ozlabs.org/project/qemu-devel/patch/159679993438.876294.7285654331498605426.st...@bahia.lan/
>
> which prevents of a bogus error path in kvmppc_xive_source_reset_one() to
> be taken:
Oops, yeah, I realized this halfway through and fixed it up, but
forgot to push out the updated version.
>
> /* The KVM XIVE device is not in use */
> if (xive->fd == -1) {
> return -ENODEV; <== this should return 0 to avoid the segfault, but
> the real issue is that kvmppc_xive_*() calls
> shouldn't be called at all when we don't have
> a KVM XIVE device. This is the purpose of the
> "spapr: Cleanups for XIVE" series.
> }
>
> I should maybe have added some Based-on: tag to make it clearer...
That would have helped, yes.
>
> David,
>
> Can you apply the series the other way around ?
>
> First :
>
> "spapr: Cleanups for XIVE"
>
> http://patchwork.ozlabs.org/project/qemu-devel/cover/159679991916.876294.8967140647442842745.st...@bahia.lan/
>
> Then :
>
> "ppc/spapr: Error handling fixes and cleanups"
>
> http://patchwork.ozlabs.org/project/qemu-devel/cover/159707843034.1489912.1082061742626355958.st...@bahia.lan/
>
> Sorry everyone for the inconvenience.
>
--
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
Re: [PATCH 01/14] spapr: Simplify error handling in spapr_phb_realize()
On Thu, 13 Aug 2020 16:57:04 -0300
Daniel Henrique Barboza wrote:
> Greg,
>
> This patch is breaking guest startup in ppc-for-5.2 for me. The process
> gives an almost instant segfault. Here's what I'm doing:
>
Ok, this is because this patch should have been applied after
the "spapr: Cleanups for XIVE" series:
http://patchwork.ozlabs.org/project/qemu-devel/cover/159679991916.876294.8967140647442842745.st...@bahia.lan/
specifically this patch:
http://patchwork.ozlabs.org/project/qemu-devel/patch/159679993438.876294.7285654331498605426.st...@bahia.lan/
which prevents of a bogus error path in kvmppc_xive_source_reset_one() to
be taken:
/* The KVM XIVE device is not in use */
if (xive->fd == -1) {
return -ENODEV; <== this should return 0 to avoid the segfault, but
the real issue is that kvmppc_xive_*() calls
shouldn't be called at all when we don't have
a KVM XIVE device. This is the purpose of the
"spapr: Cleanups for XIVE" series.
}
I should maybe have added some Based-on: tag to make it clearer...
David,
Can you apply the series the other way around ?
First :
"spapr: Cleanups for XIVE"
http://patchwork.ozlabs.org/project/qemu-devel/cover/159679991916.876294.8967140647442842745.st...@bahia.lan/
Then :
"ppc/spapr: Error handling fixes and cleanups"
http://patchwork.ozlabs.org/project/qemu-devel/cover/159707843034.1489912.1082061742626355958.st...@bahia.lan/
Sorry everyone for the inconvenience.
Cheers,
--
Greg
> $ sudo ./qemu-system-ppc64 -machine
> pseries-5.1,accel=kvm,usb=off,dump-guest-core=off -m 65536\
> -overcommit mem-lock=off -smp 4,sockets=4,cores=1,threads=1 -rtc base=utc
> -display none -vga none -nographic -boot menu=on \
> -device spapr-pci-host-bridge,index=1,id=pci.1 -device
> spapr-pci-host-bridge,index=2,id=pci.2 \
> -device spapr-pci-host-bridge,index=3,id=pci.3 -device
> spapr-pci-host-bridge,index=4,id=pci.4 \
> -device qemu-xhci,id=usb,bus=pci.0,addr=0x2 \
> -drive
> file=/home/danielhb/f32.qcow2,format=qcow2,if=none,id=drive-virtio-disk0 \
> -device
> virtio-blk-pci,scsi=off,bus=pci.0,addr=0x3,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1
> \
> -device usb-kbd,id=input0,bus=usb.0,port=1 -device
> usb-mouse,id=input1,bus=usb.0,port=2 \
> -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x4 -msg timestamp=on \
> -machine cap-ccf-assist=off
> Segmentation fault
> $
>
>
> GDB points this backtrace:
>
> Thread 1 "qemu-system-ppc" received signal SIGSEGV, Segmentation fault.
> error_vprepend (errp=0x7fffe400, fmt=0x100ec2698 "can't allocate LSIs: ",
> ap=0x7fffe290 "\030") at /home/danielhb/qemu/util/error.c:134
> 134 g_string_append(newmsg, (*errp)->msg);
> Missing separate debuginfos, use: dnf debuginfo-install
> glib2-2.64.4-1.fc32.ppc64le libblkid-2.35.2-1.fc32.ppc64le
> libffi-3.1-24.fc32.ppc64le libgcrypt-1.8.5-3.fc32.ppc64le
> libgpg-error-1.36-3.fc32.ppc64le libmount-2.35.2-1.fc32.ppc64le
> libselinux-3.0-5.fc32.ppc64le libxml2-2.9.10-3.fc32.ppc64le
> ncurses-libs-6.1-15.20191109.fc32.ppc64le numactl-libs-2.0.12-4.fc32.ppc64le
> pcre-8.44-1.fc32.ppc64le pcre2-10.35-4.fc32.ppc64le
> pixman-0.40.0-1.fc32.ppc64le xz-libs-5.2.5-1.fc32.ppc64le
> zlib-1.2.11-21.fc32.ppc64le
> (gdb) bt
> #0 error_vprepend (errp=0x7fffe400, fmt=0x100ec2698 "can't allocate
> LSIs: ", ap=0x7fffe290 "\030") at /home/danielhb/qemu/util/error.c:134
> #1 0x000100c1e9cc in error_prepend (errp=0x7fffe400, fmt=0x100ec2698
> "can't allocate LSIs: ") at /home/danielhb/qemu/util/error.c:144
> #2 0x0001004cdad4 in spapr_phb_realize (dev=0x101d6cb90,
> errp=0x7fffe400) at /home/danielhb/qemu/hw/ppc/spapr_pci.c:1982
> #3 0x000100735f70 in device_set_realized (obj=0x101d6cb90, value=true,
> errp=0x7fffe568) at /home/danielhb/qemu/hw/core/qdev.c:864
> #4 0x000100a5aae4 in property_set_bool (obj=0x101d6cb90, v=0x101d6daa0,
> name=0x100f13df8 "realized", opaque=0x1016d2430, errp=0x7fffe568) at
> /home/danielhb/qemu/qom/object.c:2202
> #5 0x000100a57d64 in object_property_set (obj=0x101d6cb90,
> name=0x100f13df8 "realized", v=0x101d6daa0, errp=0x1016156c0 )
> at /home/danielhb/qemu/qom/object.c:1349
> #6 0x000100a5cb38 in object_property_set_qobject (obj=0x101d6cb90,
> name=0x100f13df8 "realized", value=0x101d6c950, errp=0x1016156c0
> )
> at /home/danielhb/qemu/qom/qom-qobject.c:28
> #7 0x000100a581fc in object_property_set_bool (obj=0x101d6cb90,
> name=0x100f13df8 "realized", value=true, errp=0x1016156c0 ) at
> /home/danielhb/qemu/qom/object.c:1416
> #8 0x000100734178 in qdev_realize (dev=0x101d6cb90, bus=0x10198e250,
> errp=0x1016156c0 ) at /home/danielhb/qemu/hw/core/qdev.c:379
> #9 0x0001007341dc in qdev_realize_and_unref (dev=0x101d6cb90,
> bus=0x10198e250, errp=0x1016156c0 ) at
> /home/danielhb/qemu/hw/core/qdev.c:386
> #10 0x0001007463c0 in
Re: [PATCH 01/14] spapr: Simplify error handling in spapr_phb_realize()
Greg, This patch is breaking guest startup in ppc-for-5.2 for me. The process gives an almost instant segfault. Here's what I'm doing: $ sudo ./qemu-system-ppc64 -machine pseries-5.1,accel=kvm,usb=off,dump-guest-core=off -m 65536\ -overcommit mem-lock=off -smp 4,sockets=4,cores=1,threads=1 -rtc base=utc -display none -vga none -nographic -boot menu=on \ -device spapr-pci-host-bridge,index=1,id=pci.1 -device spapr-pci-host-bridge,index=2,id=pci.2 \ -device spapr-pci-host-bridge,index=3,id=pci.3 -device spapr-pci-host-bridge,index=4,id=pci.4 \ -device qemu-xhci,id=usb,bus=pci.0,addr=0x2 \ -drive file=/home/danielhb/f32.qcow2,format=qcow2,if=none,id=drive-virtio-disk0 \ -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x3,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1 \ -device usb-kbd,id=input0,bus=usb.0,port=1 -device usb-mouse,id=input1,bus=usb.0,port=2 \ -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x4 -msg timestamp=on \ -machine cap-ccf-assist=off Segmentation fault $ GDB points this backtrace: Thread 1 "qemu-system-ppc" received signal SIGSEGV, Segmentation fault. error_vprepend (errp=0x7fffe400, fmt=0x100ec2698 "can't allocate LSIs: ", ap=0x7fffe290 "\030") at /home/danielhb/qemu/util/error.c:134 134 g_string_append(newmsg, (*errp)->msg); Missing separate debuginfos, use: dnf debuginfo-install glib2-2.64.4-1.fc32.ppc64le libblkid-2.35.2-1.fc32.ppc64le libffi-3.1-24.fc32.ppc64le libgcrypt-1.8.5-3.fc32.ppc64le libgpg-error-1.36-3.fc32.ppc64le libmount-2.35.2-1.fc32.ppc64le libselinux-3.0-5.fc32.ppc64le libxml2-2.9.10-3.fc32.ppc64le ncurses-libs-6.1-15.20191109.fc32.ppc64le numactl-libs-2.0.12-4.fc32.ppc64le pcre-8.44-1.fc32.ppc64le pcre2-10.35-4.fc32.ppc64le pixman-0.40.0-1.fc32.ppc64le xz-libs-5.2.5-1.fc32.ppc64le zlib-1.2.11-21.fc32.ppc64le (gdb) bt #0 error_vprepend (errp=0x7fffe400, fmt=0x100ec2698 "can't allocate LSIs: ", ap=0x7fffe290 "\030") at /home/danielhb/qemu/util/error.c:134 #1 0x000100c1e9cc in error_prepend (errp=0x7fffe400, fmt=0x100ec2698 "can't allocate LSIs: ") at /home/danielhb/qemu/util/error.c:144 #2 0x0001004cdad4 in spapr_phb_realize (dev=0x101d6cb90, errp=0x7fffe400) at /home/danielhb/qemu/hw/ppc/spapr_pci.c:1982 #3 0x000100735f70 in device_set_realized (obj=0x101d6cb90, value=true, errp=0x7fffe568) at /home/danielhb/qemu/hw/core/qdev.c:864 #4 0x000100a5aae4 in property_set_bool (obj=0x101d6cb90, v=0x101d6daa0, name=0x100f13df8 "realized", opaque=0x1016d2430, errp=0x7fffe568) at /home/danielhb/qemu/qom/object.c:2202 #5 0x000100a57d64 in object_property_set (obj=0x101d6cb90, name=0x100f13df8 "realized", v=0x101d6daa0, errp=0x1016156c0 ) at /home/danielhb/qemu/qom/object.c:1349 #6 0x000100a5cb38 in object_property_set_qobject (obj=0x101d6cb90, name=0x100f13df8 "realized", value=0x101d6c950, errp=0x1016156c0 ) at /home/danielhb/qemu/qom/qom-qobject.c:28 #7 0x000100a581fc in object_property_set_bool (obj=0x101d6cb90, name=0x100f13df8 "realized", value=true, errp=0x1016156c0 ) at /home/danielhb/qemu/qom/object.c:1416 #8 0x000100734178 in qdev_realize (dev=0x101d6cb90, bus=0x10198e250, errp=0x1016156c0 ) at /home/danielhb/qemu/hw/core/qdev.c:379 #9 0x0001007341dc in qdev_realize_and_unref (dev=0x101d6cb90, bus=0x10198e250, errp=0x1016156c0 ) at /home/danielhb/qemu/hw/core/qdev.c:386 #10 0x0001007463c0 in sysbus_realize_and_unref (dev=0x101d6cb90, errp=0x1016156c0 ) at /home/danielhb/qemu/hw/core/sysbus.c:260 #11 0x0001004a6960 in spapr_create_default_phb () at /home/danielhb/qemu/hw/ppc/spapr.c:2652 #12 0x0001004a7428 in spapr_machine_init (machine=0x101965800) at /home/danielhb/qemu/hw/ppc/spapr.c:2940 #13 0x00010074a3b0 in machine_run_board_init (machine=0x101965800) at /home/danielhb/qemu/hw/core/machine.c:1135 #14 0x00010054f390 in qemu_init (argc=42, argv=0x70a8, envp=0x7200) at /home/danielhb/qemu/softmmu/vl.c:4355 #15 0x000100b8ee00 in main (argc=42, argv=0x70a8, envp=0x7200) at /home/danielhb/qemu/softmmu/main.c:48 (gdb) Removing this patch (i.e. resetting HEAD at "target/ppc: Integrate icount to purr, vtb, and tbu40") allows me to get the guest rolling. Thanks, Daniel On 8/10/20 1:53 PM, Greg Kurz wrote: The spapr_phb_realize() function has a local_err variable which is used to: 1) check failures of spapr_irq_findone() and spapr_irq_claim() 2) prepend extra information to the error message Recent work from Markus Armbruster highlighted we get better code when testing the return value of a function, rather than setting up all the local_err boiler plate. For similar reasons, it is now preferred to use ERRP_GUARD() and error_prepend() rather than error_propagate_prepend(). Since spapr_irq_findone() and spapr_irq_claim() return negative values in case of failure, do both changes. This is just cleanup, no functional impact. Signed-off-by: Greg Kurz Reviewed-by:
Re: [PATCH 01/14] spapr: Simplify error handling in spapr_phb_realize()
On Mon, Aug 10, 2020 at 06:53:58PM +0200, Greg Kurz wrote:
> The spapr_phb_realize() function has a local_err variable which
> is used to:
>
> 1) check failures of spapr_irq_findone() and spapr_irq_claim()
>
> 2) prepend extra information to the error message
>
> Recent work from Markus Armbruster highlighted we get better
> code when testing the return value of a function, rather than
> setting up all the local_err boiler plate. For similar reasons,
> it is now preferred to use ERRP_GUARD() and error_prepend()
> rather than error_propagate_prepend().
>
> Since spapr_irq_findone() and spapr_irq_claim() return negative
> values in case of failure, do both changes.
>
> This is just cleanup, no functional impact.
>
> Signed-off-by: Greg Kurz
> Reviewed-by: Markus Armbruster
> Reviewed-by: David Gibson
Applied to ppc-for-5.2.
> ---
> hw/ppc/spapr_pci.c | 16 +++-
> 1 file changed, 7 insertions(+), 9 deletions(-)
>
> diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
> index 363cdb3f7b8d..0a418f1e6711 100644
> --- a/hw/ppc/spapr_pci.c
> +++ b/hw/ppc/spapr_pci.c
> @@ -1796,6 +1796,7 @@ static void spapr_phb_destroy_msi(gpointer opaque)
>
> static void spapr_phb_realize(DeviceState *dev, Error **errp)
> {
> +ERRP_GUARD();
> /* We don't use SPAPR_MACHINE() in order to exit gracefully if the user
> * tries to add a sPAPR PHB to a non-pseries machine.
> */
> @@ -1813,7 +1814,6 @@ static void spapr_phb_realize(DeviceState *dev, Error
> **errp)
> uint64_t msi_window_size = 4096;
> SpaprTceTable *tcet;
> const unsigned windows_supported = spapr_phb_windows_supported(sphb);
> -Error *local_err = NULL;
>
> if (!spapr) {
> error_setg(errp, TYPE_SPAPR_PCI_HOST_BRIDGE " needs a pseries
> machine");
> @@ -1964,13 +1964,12 @@ static void spapr_phb_realize(DeviceState *dev, Error
> **errp)
>
> /* Initialize the LSI table */
> for (i = 0; i < PCI_NUM_PINS; i++) {
> -uint32_t irq = SPAPR_IRQ_PCI_LSI + sphb->index * PCI_NUM_PINS + i;
> +int irq = SPAPR_IRQ_PCI_LSI + sphb->index * PCI_NUM_PINS + i;
>
> if (smc->legacy_irq_allocation) {
> -irq = spapr_irq_findone(spapr, _err);
> -if (local_err) {
> -error_propagate_prepend(errp, local_err,
> -"can't allocate LSIs: ");
> +irq = spapr_irq_findone(spapr, errp);
> +if (irq < 0) {
> +error_prepend(errp, "can't allocate LSIs: ");
> /*
> * Older machines will never support PHB hotplug, ie, this
> is an
> * init only path and QEMU will terminate. No need to
> rollback.
> @@ -1979,9 +1978,8 @@ static void spapr_phb_realize(DeviceState *dev, Error
> **errp)
> }
> }
>
> -spapr_irq_claim(spapr, irq, true, _err);
> -if (local_err) {
> -error_propagate_prepend(errp, local_err, "can't allocate LSIs:
> ");
> +if (spapr_irq_claim(spapr, irq, true, errp) < 0) {
> +error_prepend(errp, "can't allocate LSIs: ");
> goto unrealize;
> }
>
>
>
--
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
[PATCH 01/14] spapr: Simplify error handling in spapr_phb_realize()
The spapr_phb_realize() function has a local_err variable which
is used to:
1) check failures of spapr_irq_findone() and spapr_irq_claim()
2) prepend extra information to the error message
Recent work from Markus Armbruster highlighted we get better
code when testing the return value of a function, rather than
setting up all the local_err boiler plate. For similar reasons,
it is now preferred to use ERRP_GUARD() and error_prepend()
rather than error_propagate_prepend().
Since spapr_irq_findone() and spapr_irq_claim() return negative
values in case of failure, do both changes.
This is just cleanup, no functional impact.
Signed-off-by: Greg Kurz
Reviewed-by: Markus Armbruster
Reviewed-by: David Gibson
---
hw/ppc/spapr_pci.c | 16 +++-
1 file changed, 7 insertions(+), 9 deletions(-)
diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index 363cdb3f7b8d..0a418f1e6711 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -1796,6 +1796,7 @@ static void spapr_phb_destroy_msi(gpointer opaque)
static void spapr_phb_realize(DeviceState *dev, Error **errp)
{
+ERRP_GUARD();
/* We don't use SPAPR_MACHINE() in order to exit gracefully if the user
* tries to add a sPAPR PHB to a non-pseries machine.
*/
@@ -1813,7 +1814,6 @@ static void spapr_phb_realize(DeviceState *dev, Error
**errp)
uint64_t msi_window_size = 4096;
SpaprTceTable *tcet;
const unsigned windows_supported = spapr_phb_windows_supported(sphb);
-Error *local_err = NULL;
if (!spapr) {
error_setg(errp, TYPE_SPAPR_PCI_HOST_BRIDGE " needs a pseries
machine");
@@ -1964,13 +1964,12 @@ static void spapr_phb_realize(DeviceState *dev, Error
**errp)
/* Initialize the LSI table */
for (i = 0; i < PCI_NUM_PINS; i++) {
-uint32_t irq = SPAPR_IRQ_PCI_LSI + sphb->index * PCI_NUM_PINS + i;
+int irq = SPAPR_IRQ_PCI_LSI + sphb->index * PCI_NUM_PINS + i;
if (smc->legacy_irq_allocation) {
-irq = spapr_irq_findone(spapr, _err);
-if (local_err) {
-error_propagate_prepend(errp, local_err,
-"can't allocate LSIs: ");
+irq = spapr_irq_findone(spapr, errp);
+if (irq < 0) {
+error_prepend(errp, "can't allocate LSIs: ");
/*
* Older machines will never support PHB hotplug, ie, this is
an
* init only path and QEMU will terminate. No need to rollback.
@@ -1979,9 +1978,8 @@ static void spapr_phb_realize(DeviceState *dev, Error
**errp)
}
}
-spapr_irq_claim(spapr, irq, true, _err);
-if (local_err) {
-error_propagate_prepend(errp, local_err, "can't allocate LSIs: ");
+if (spapr_irq_claim(spapr, irq, true, errp) < 0) {
+error_prepend(errp, "can't allocate LSIs: ");
goto unrealize;
}
