From: Marc Zyngier <maz@kernel.org> To: Zenghui Yu <yuzenghui@huawei.com> Cc: jason@lakedaemon.net, jiayanlei@huawei.com, tglx@linutronix.de, kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org Subject: Re: [PATCH] irqchip/gic-v3-its: Use the exact ITSList for VMOVP Date: Tue, 22 Oct 2019 13:44:46 +0100 [thread overview] Message-ID: <34e9236f057b22d49f40146b21f81282@www.loen.fr> (raw) In-Reply-To: <1571742366-21008-1-git-send-email-yuzenghui@huawei.com> Hi Zenghui, On 2019-10-22 12:06, Zenghui Yu wrote: > On a system without Single VMOVP support (say GITS_TYPER.VMOVP == 0), > we will map vPEs only on ITSs that will actually control interrupts > for the given VM. And when moving a vPE, the VMOVP command will be > issued only for those ITSs. > > But when issuing VMOVPs we seemed fail to present the exact ITSList > to ITSs who are actually included in the synchronization operation. > The its_list_map we're currently using includes all ITSs in the > system, > even though some of them don't have the corrsponding vPE mapping at > all. > > Introduce get_its_list() to get the per-VM its_list_map, to indicate > which ITSs have vPE mappings for the given VM, and use this map as > the expected ITSList when building VMOVP. > > Signed-off-by: Zenghui Yu <yuzenghui@huawei.com> > --- > > I haven't seen any broken with the current its_list_map, but behavior > might differ between implementations. Let's do what the spec expects > us to do and try not to confuse the implementation. Or is there any > points I've missed? No, I think you're right, and this is just an oversight on my part: for example, we seem to do the right thing when handling a guest invall, where we scan the ITS nodes and only emit a vinvall on an ITS that has VLPI mapped in. I don't think this is likely to cause any harm (after all, a DISCARD and a VMOVP could race at any time), but it is certainly a performance gain not to throw extra commands at unsuspecting ITSs. So thanks for spotting this. A couple of comments below: > > drivers/irqchip/irq-gic-v3-its.c | 15 ++++++++++++++- > 1 file changed, 14 insertions(+), 1 deletion(-) > > diff --git a/drivers/irqchip/irq-gic-v3-its.c > b/drivers/irqchip/irq-gic-v3-its.c > index c81da27044bf..eebee588ea30 100644 > --- a/drivers/irqchip/irq-gic-v3-its.c > +++ b/drivers/irqchip/irq-gic-v3-its.c > @@ -175,6 +175,19 @@ static DEFINE_IDA(its_vpeid_ida); > #define gic_data_rdist_rd_base() (gic_data_rdist()->rd_base) > #define gic_data_rdist_vlpi_base() (gic_data_rdist_rd_base() + > SZ_128K) > > +static inline u16 get_its_list(struct its_vm *vm) Please drop the inline, the compiler will do it for you. > +{ > + struct its_node *its; > + unsigned long its_list; > + > + list_for_each_entry(its, &its_nodes, entry) { You probably want to skip non v4 ITSs, as they don't have a list_nr associated to them (and you'd probably end-up hitting ITS #0 for no good reason). > + if (vm->vlpi_count[its->list_nr]) > + set_bit(its->list_nr, &its_list); We don't need to be atomic here, so __set_bit would be just as fine. > + } > + > + return (u16)its_list; > +} > + > static struct its_collection *dev_event_to_col(struct its_device > *its_dev, > u32 event) > { > @@ -982,7 +995,6 @@ static void its_send_vmovp(struct its_vpe *vpe) > int col_id = vpe->col_idx; > > desc.its_vmovp_cmd.vpe = vpe; > - desc.its_vmovp_cmd.its_list = (u16)its_list_map; Careful here, you're leaving the its_list field uninitialized, and it really should be 0 when GITS_TYPER.VMOVP == 1 (i.e. when its_list_map is zero). Just initialize the whole descriptor to zero. > > if (!its_list_map) { > its = list_first_entry(&its_nodes, struct its_node, entry); > @@ -1003,6 +1015,7 @@ static void its_send_vmovp(struct its_vpe *vpe) > raw_spin_lock_irqsave(&vmovp_lock, flags); > > desc.its_vmovp_cmd.seq_num = vmovp_seq_num++; > + desc.its_vmovp_cmd.its_list = get_its_list(vpe->its_vm); > > /* Emit VMOVPs */ > list_for_each_entry(its, &its_nodes, entry) { Thanks, M. -- Jazz is not dead. It just smells funny... _______________________________________________ kvmarm mailing list kvmarm@lists.cs.columbia.edu https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
WARNING: multiple messages have this Message-ID (diff)
From: Marc Zyngier <maz@kernel.org> To: Zenghui Yu <yuzenghui@huawei.com> Cc: jason@lakedaemon.net, jiayanlei@huawei.com, wanghaibin.wang@huawei.com, tglx@linutronix.de, kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org Subject: Re: [PATCH] irqchip/gic-v3-its: Use the exact ITSList for VMOVP Date: Tue, 22 Oct 2019 13:44:46 +0100 [thread overview] Message-ID: <34e9236f057b22d49f40146b21f81282@www.loen.fr> (raw) In-Reply-To: <1571742366-21008-1-git-send-email-yuzenghui@huawei.com> Hi Zenghui, On 2019-10-22 12:06, Zenghui Yu wrote: > On a system without Single VMOVP support (say GITS_TYPER.VMOVP == 0), > we will map vPEs only on ITSs that will actually control interrupts > for the given VM. And when moving a vPE, the VMOVP command will be > issued only for those ITSs. > > But when issuing VMOVPs we seemed fail to present the exact ITSList > to ITSs who are actually included in the synchronization operation. > The its_list_map we're currently using includes all ITSs in the > system, > even though some of them don't have the corrsponding vPE mapping at > all. > > Introduce get_its_list() to get the per-VM its_list_map, to indicate > which ITSs have vPE mappings for the given VM, and use this map as > the expected ITSList when building VMOVP. > > Signed-off-by: Zenghui Yu <yuzenghui@huawei.com> > --- > > I haven't seen any broken with the current its_list_map, but behavior > might differ between implementations. Let's do what the spec expects > us to do and try not to confuse the implementation. Or is there any > points I've missed? No, I think you're right, and this is just an oversight on my part: for example, we seem to do the right thing when handling a guest invall, where we scan the ITS nodes and only emit a vinvall on an ITS that has VLPI mapped in. I don't think this is likely to cause any harm (after all, a DISCARD and a VMOVP could race at any time), but it is certainly a performance gain not to throw extra commands at unsuspecting ITSs. So thanks for spotting this. A couple of comments below: > > drivers/irqchip/irq-gic-v3-its.c | 15 ++++++++++++++- > 1 file changed, 14 insertions(+), 1 deletion(-) > > diff --git a/drivers/irqchip/irq-gic-v3-its.c > b/drivers/irqchip/irq-gic-v3-its.c > index c81da27044bf..eebee588ea30 100644 > --- a/drivers/irqchip/irq-gic-v3-its.c > +++ b/drivers/irqchip/irq-gic-v3-its.c > @@ -175,6 +175,19 @@ static DEFINE_IDA(its_vpeid_ida); > #define gic_data_rdist_rd_base() (gic_data_rdist()->rd_base) > #define gic_data_rdist_vlpi_base() (gic_data_rdist_rd_base() + > SZ_128K) > > +static inline u16 get_its_list(struct its_vm *vm) Please drop the inline, the compiler will do it for you. > +{ > + struct its_node *its; > + unsigned long its_list; > + > + list_for_each_entry(its, &its_nodes, entry) { You probably want to skip non v4 ITSs, as they don't have a list_nr associated to them (and you'd probably end-up hitting ITS #0 for no good reason). > + if (vm->vlpi_count[its->list_nr]) > + set_bit(its->list_nr, &its_list); We don't need to be atomic here, so __set_bit would be just as fine. > + } > + > + return (u16)its_list; > +} > + > static struct its_collection *dev_event_to_col(struct its_device > *its_dev, > u32 event) > { > @@ -982,7 +995,6 @@ static void its_send_vmovp(struct its_vpe *vpe) > int col_id = vpe->col_idx; > > desc.its_vmovp_cmd.vpe = vpe; > - desc.its_vmovp_cmd.its_list = (u16)its_list_map; Careful here, you're leaving the its_list field uninitialized, and it really should be 0 when GITS_TYPER.VMOVP == 1 (i.e. when its_list_map is zero). Just initialize the whole descriptor to zero. > > if (!its_list_map) { > its = list_first_entry(&its_nodes, struct its_node, entry); > @@ -1003,6 +1015,7 @@ static void its_send_vmovp(struct its_vpe *vpe) > raw_spin_lock_irqsave(&vmovp_lock, flags); > > desc.its_vmovp_cmd.seq_num = vmovp_seq_num++; > + desc.its_vmovp_cmd.its_list = get_its_list(vpe->its_vm); > > /* Emit VMOVPs */ > list_for_each_entry(its, &its_nodes, entry) { Thanks, M. -- Jazz is not dead. It just smells funny... _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2019-10-22 12:44 UTC|newest] Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top 2019-10-22 11:06 [PATCH] irqchip/gic-v3-its: Use the exact ITSList for VMOVP Zenghui Yu 2019-10-22 11:06 ` Zenghui Yu 2019-10-22 12:44 ` Marc Zyngier [this message] 2019-10-22 12:44 ` Marc Zyngier 2019-10-23 2:08 ` Zenghui Yu 2019-10-23 2:08 ` Zenghui Yu
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=34e9236f057b22d49f40146b21f81282@www.loen.fr \ --to=maz@kernel.org \ --cc=jason@lakedaemon.net \ --cc=jiayanlei@huawei.com \ --cc=kvmarm@lists.cs.columbia.edu \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=tglx@linutronix.de \ --cc=yuzenghui@huawei.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.