* [PATCH v4 00/23] Xenstore stub domain
@ 2012-01-25 14:37 Daniel De Graaf
2012-01-25 14:37 ` [PATCH 01/23] xen: reinstate previously unused XENMEM_remove_from_physmap hypercall Daniel De Graaf
` (22 more replies)
0 siblings, 23 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel
Changes from v3:
- mini-os configuration files moved into stubdom/
- mini-os extra console support now a config option
- Fewer #ifdefs
- grant table setup uses hypercall bounce
- Xenstore stub domain syslog support re-enabled
Changes from v2:
- configuration support added to mini-os build system
- add mini-os support for conditionally compiling frontends, xenbus
- XENMEM_remove_from_physmap moved out of arch-specific code
- use uint32_t for virqs
- warn when dropping grant v2-only flags when switching versions
- IOCTL_XENBUS_BACKEND_SETUP name changed so userspace can implement compat
- ioctl now returns -EEXIST if xenstored has already been connected
- various cosmetic cleanups, shuffling
Changes from v1:
- set_virq_handler implemented in libxc
- added custom domain builder for xenstored
- xenstore/console domain IDs now pulled from xenstore
- migration support when using split xenstored (untested, should work)
- slightly less intrusive NO_SOCKETS xenstored patch
(still has many ifdefs to avoid pulling in socket headers or symbols)
- virq handlers removed from dying domain when clearing event channels
- dummy XSM module restricts getdomaininfo similar to no-XSM case
- formatting/type fixups
- partial ioctl compatibility with legacy IOCTL_XENBUS_ALLOC
To start xenstored, run:
tools/xenstore/init-xenstore-domain stubdom/mini-os-x86_64-xenstore/mini-os 20 system_u:system_r:domU_t
This will populate the xenstore domid key /tool/xenstore/domid
Other notes:
The console for xenstored is not currently set up by
init-xenstore-domain. If the hypervisor is compiled with VERBOSE or
debug=y, output from xenstored will be visible on the hypervisor serial
console (or ring buffer if enabled with console_to_ring). The xenstore
stub domain itself supports console output, and init-xenstore-domain
could be extended to daemonize and spool this output to a log file. The
normal xenconsole daemon cannot be used here due to the possibility of a
deadlock.
----
[PATCH 01/23] xen: reinstate previously unused
[PATCH 02/23] xen: allow global VIRQ handlers to be delegated to
[PATCH 03/23] xen: change virq parameters from int to uint32_t
[PATCH 04/23] xen: use XSM instead of IS_PRIV for getdomaininfo
[PATCH 05/23] xen: Preserve reserved grant entries when switching
[PATCH 06/23] tools/libxl: pull xenstore/console domids from
[PATCH 07/23] lib{xc,xl}: Seed grant tables with xenstore and
[PATCH 08/23] mini-os: avoid crash if no console is provided
[PATCH 09/23] mini-os: remove per-fd evtchn limit
[PATCH 10/23] mini-os: create app-specific configuration
[PATCH 11/23] mini-os: Move test functions into test.c
[PATCH 12/23] mini-os: make frontends and xenbus optional
[PATCH 13/23] mini-os: fix list.h include guard name
[PATCH 14/23] xenstored: use grant references instead of
[PATCH 15/23] xenstored: refactor socket setup code
[PATCH 16/23] xenstored: add NO_SOCKETS compilation option
[PATCH 17/23] xenstored: support for tdb_copy with TDB_INTERNAL
[PATCH 18/23] xenstored: support running in minios stubdom
[PATCH 19/23] stubdom: enable xenstored build
[PATCH 20/23] xenstored: add --event parameter for bootstrapping
[PATCH 21/23] xenstored: use domain_is_unprivileged instead of
[PATCH 22/23] xenstored: add --priv-domid parameter
[PATCH 23/23] xenstored: Add stub domain builder
Linux patch unchanged since v3, not reposted this time:
[PATCH] xenbus: Add support for xenbus backend in stub domain
^ permalink raw reply [flat|nested] 27+ messages in thread
* [PATCH 01/23] xen: reinstate previously unused XENMEM_remove_from_physmap hypercall
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 02/23] xen: allow global VIRQ handlers to be delegated to other domains Daniel De Graaf
` (21 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
This patch reinstates the XENMEM_remove_from_physmap hypercall
which was removed in 19041:ee62aaafff46 because it was not used.
However, is now needed in order to support xenstored stub domains.
The xenstored stub domain is not priviliged like dom0 and so cannot
unilaterally map the xenbus page of other guests into it's address
space. Therefore, before creating a domU the domain builder needs to
seed its grant table with a grant ref allowing the xenstored stub
domain to access the new domU's xenbus page.
At present domU's do not start with their grant table mapped.
Instead it gets mapped when the guest requests a grant table from
the hypervisor.
In order to seed the grant table, the domain builder first needs to
map it into dom0 address space. But the hypercall to do this
requires a gpfn (guest pfn), which is an mfn for PV guest, but a pfn
for HVM guests. Therfore, in order to seed the grant table of an
HVM guest, dom0 needs to *temporarily* map it into the guest's
"physical" address space.
Hence the need to reinstate the XENMEM_remove_from_physmap hypercall.
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
xen/common/compat/memory.c | 14 ++++++++++++++
xen/common/memory.c | 37 +++++++++++++++++++++++++++++++++++++
xen/include/asm-ia64/mm.h | 1 +
xen/include/public/memory.h | 16 ++++++++++++++++
xen/include/xlat.lst | 1 +
xen/include/xsm/xsm.h | 6 ++++++
xen/xsm/dummy.c | 6 ++++++
xen/xsm/flask/hooks.c | 6 ++++++
8 files changed, 87 insertions(+), 0 deletions(-)
diff --git a/xen/common/compat/memory.c b/xen/common/compat/memory.c
index 2402984..e7257cc 100644
--- a/xen/common/compat/memory.c
+++ b/xen/common/compat/memory.c
@@ -25,6 +25,7 @@ int compat_memory_op(unsigned int cmd, XEN_GUEST_HANDLE(void) compat)
XEN_GUEST_HANDLE(void) hnd;
struct xen_memory_reservation *rsrv;
struct xen_memory_exchange *xchg;
+ struct xen_remove_from_physmap *xrfp;
} nat;
union {
struct compat_memory_reservation rsrv;
@@ -179,6 +180,18 @@ int compat_memory_op(unsigned int cmd, XEN_GUEST_HANDLE(void) compat)
nat.hnd = compat;
break;
+ case XENMEM_remove_from_physmap:
+ {
+ struct compat_remove_from_physmap cmp;
+
+ if ( copy_from_guest(&cmp, compat, 1) )
+ return -EFAULT;
+
+ XLAT_remove_from_physmap(nat.xrfp, &cmp);
+
+ break;
+ }
+
default:
return compat_arch_memory_op(cmd, compat);
}
@@ -284,6 +297,7 @@ int compat_memory_op(unsigned int cmd, XEN_GUEST_HANDLE(void) compat)
case XENMEM_current_reservation:
case XENMEM_maximum_reservation:
case XENMEM_maximum_gpfn:
+ case XENMEM_remove_from_physmap:
break;
default:
diff --git a/xen/common/memory.c b/xen/common/memory.c
index 8d45439..53886ce 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -659,6 +659,43 @@ long do_memory_op(unsigned long cmd, XEN_GUEST_HANDLE(void) arg)
break;
+ case XENMEM_remove_from_physmap:
+ {
+ struct xen_remove_from_physmap xrfp;
+ unsigned long mfn;
+ struct domain *d;
+
+ if ( copy_from_guest(&xrfp, arg, 1) )
+ return -EFAULT;
+
+ rc = rcu_lock_target_domain_by_id(xrfp.domid, &d);
+ if ( rc != 0 )
+ return rc;
+
+ if ( xsm_remove_from_physmap(current->domain, d) )
+ {
+ rcu_unlock_domain(d);
+ return -EPERM;
+ }
+
+ domain_lock(d);
+
+ mfn = get_gfn_untyped(d, xrfp.gpfn);
+
+ if ( mfn_valid(mfn) )
+ guest_physmap_remove_page(d, xrfp.gpfn, mfn, PAGE_ORDER_4K);
+ else
+ rc = -ENOENT;
+
+ put_gfn(d, xrfp.gpfn);
+
+ domain_unlock(d);
+
+ rcu_unlock_domain(d);
+
+ break;
+ }
+
default:
rc = arch_memory_op(op, arg);
break;
diff --git a/xen/include/asm-ia64/mm.h b/xen/include/asm-ia64/mm.h
index d09c363..a2bfc02 100644
--- a/xen/include/asm-ia64/mm.h
+++ b/xen/include/asm-ia64/mm.h
@@ -550,6 +550,7 @@ extern u64 translate_domain_pte(u64 pteval, u64 address, u64 itir__,
#define gmfn_to_mfn(_d, gpfn) \
gmfn_to_mfn_foreign((_d), (gpfn))
+#define get_gfn_untyped(d, gpfn) gmfn_to_mfn(d, gpfn)
#define put_gfn(d, g) ((void)0)
#define __gpfn_invalid(_d, gpfn) \
diff --git a/xen/include/public/memory.h b/xen/include/public/memory.h
index c5b78a8..308deff 100644
--- a/xen/include/public/memory.h
+++ b/xen/include/public/memory.h
@@ -229,6 +229,22 @@ struct xen_add_to_physmap {
typedef struct xen_add_to_physmap xen_add_to_physmap_t;
DEFINE_XEN_GUEST_HANDLE(xen_add_to_physmap_t);
+/*
+ * Unmaps the page appearing at a particular GPFN from the specified guest's
+ * pseudophysical address space.
+ * arg == addr of xen_remove_from_physmap_t.
+ */
+#define XENMEM_remove_from_physmap 15
+struct xen_remove_from_physmap {
+ /* Which domain to change the mapping for. */
+ domid_t domid;
+
+ /* GPFN of the current mapping of the page. */
+ xen_pfn_t gpfn;
+};
+typedef struct xen_remove_from_physmap xen_remove_from_physmap_t;
+DEFINE_XEN_GUEST_HANDLE(xen_remove_from_physmap_t);
+
/*** REMOVED ***/
/*#define XENMEM_translate_gpfn_list 8*/
diff --git a/xen/include/xlat.lst b/xen/include/xlat.lst
index f602155..3d4f1e3 100644
--- a/xen/include/xlat.lst
+++ b/xen/include/xlat.lst
@@ -60,6 +60,7 @@
! memory_map memory.h
! memory_reservation memory.h
! pod_target memory.h
+! remove_from_physmap memory.h
? physdev_eoi physdev.h
? physdev_get_free_pirq physdev.h
? physdev_irq physdev.h
diff --git a/xen/include/xsm/xsm.h b/xen/include/xsm/xsm.h
index df6cec2..566c808 100644
--- a/xen/include/xsm/xsm.h
+++ b/xen/include/xsm/xsm.h
@@ -169,6 +169,7 @@ struct xsm_operations {
int (*update_va_mapping) (struct domain *d, struct domain *f,
l1_pgentry_t pte);
int (*add_to_physmap) (struct domain *d1, struct domain *d2);
+ int (*remove_from_physmap) (struct domain *d1, struct domain *d2);
int (*sendtrigger) (struct domain *d);
int (*bind_pt_irq) (struct domain *d, struct xen_domctl_bind_pt_irq *bind);
int (*unbind_pt_irq) (struct domain *d);
@@ -738,6 +739,11 @@ static inline int xsm_add_to_physmap(struct domain *d1, struct domain *d2)
return xsm_call(add_to_physmap(d1, d2));
}
+static inline int xsm_remove_from_physmap(struct domain *d1, struct domain *d2)
+{
+ return xsm_call(remove_from_physmap(d1, d2));
+}
+
static inline int xsm_sendtrigger(struct domain *d)
{
return xsm_call(sendtrigger(d));
diff --git a/xen/xsm/dummy.c b/xen/xsm/dummy.c
index 4bbfbff..65daa4e 100644
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -529,6 +529,11 @@ static int dummy_add_to_physmap (struct domain *d1, struct domain *d2)
return 0;
}
+static int dummy_remove_from_physmap (struct domain *d1, struct domain *d2)
+{
+ return 0;
+}
+
static int dummy_sendtrigger (struct domain *d)
{
return 0;
@@ -690,6 +695,7 @@ void xsm_fixup_ops (struct xsm_operations *ops)
set_to_dummy_if_null(ops, mmu_machphys_update);
set_to_dummy_if_null(ops, update_va_mapping);
set_to_dummy_if_null(ops, add_to_physmap);
+ set_to_dummy_if_null(ops, remove_from_physmap);
set_to_dummy_if_null(ops, sendtrigger);
set_to_dummy_if_null(ops, bind_pt_irq);
set_to_dummy_if_null(ops, pin_mem_cacheattr);
diff --git a/xen/xsm/flask/hooks.c b/xen/xsm/flask/hooks.c
index 0d35767..a2020a9 100644
--- a/xen/xsm/flask/hooks.c
+++ b/xen/xsm/flask/hooks.c
@@ -1283,6 +1283,11 @@ static int flask_add_to_physmap(struct domain *d1, struct domain *d2)
return domain_has_perm(d1, d2, SECCLASS_MMU, MMU__PHYSMAP);
}
+static int flask_remove_from_physmap(struct domain *d1, struct domain *d2)
+{
+ return domain_has_perm(d1, d2, SECCLASS_MMU, MMU__PHYSMAP);
+}
+
static int flask_sendtrigger(struct domain *d)
{
return domain_has_perm(current->domain, d, SECCLASS_DOMAIN, DOMAIN__TRIGGER);
@@ -1550,6 +1555,7 @@ static struct xsm_operations flask_ops = {
.mmu_machphys_update = flask_mmu_machphys_update,
.update_va_mapping = flask_update_va_mapping,
.add_to_physmap = flask_add_to_physmap,
+ .remove_from_physmap = flask_remove_from_physmap,
.sendtrigger = flask_sendtrigger,
.get_device_group = flask_get_device_group,
.test_assign_device = flask_test_assign_device,
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 02/23] xen: allow global VIRQ handlers to be delegated to other domains
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
2012-01-25 14:37 ` [PATCH 01/23] xen: reinstate previously unused XENMEM_remove_from_physmap hypercall Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 03/23] xen: change virq parameters from int to uint32_t Daniel De Graaf
` (20 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
This patch sends global VIRQs to a domain designated as the VIRQ handler
instead of sending all global VIRQ events to dom0. This is required in
order to run xenstored in a stubdom, because VIRQ_DOM_EXC must be sent
to xenstored for domain destruction to work properly.
This patch was inspired by the xenstored stubdomain patch series sent to
xen-devel by Alex Zeffertt in 2009.
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/flask/policy/policy/flask/access_vectors | 1 +
tools/libxc/xc_domain.c | 10 ++++
tools/libxc/xenctrl.h | 9 +++
xen/arch/x86/cpu/mcheck/amd_nonfatal.c | 2 +-
xen/arch/x86/cpu/mcheck/mce.c | 2 +-
xen/arch/x86/cpu/mcheck/mce_intel.c | 6 +-
xen/arch/x86/cpu/mcheck/non-fatal.c | 2 +-
xen/common/cpu.c | 4 +-
xen/common/domain.c | 8 ++--
xen/common/domctl.c | 17 ++++++
xen/common/event_channel.c | 66 +++++++++++++++++++++++-
xen/common/trace.c | 2 +-
xen/drivers/char/console.c | 4 +-
xen/include/public/domctl.h | 8 +++
xen/include/xen/event.h | 12 +++-
xen/include/xsm/xsm.h | 6 ++
xen/xsm/dummy.c | 6 ++
xen/xsm/flask/hooks.c | 6 ++
xen/xsm/flask/include/av_perm_to_string.h | 1 +
xen/xsm/flask/include/av_permissions.h | 1 +
20 files changed, 154 insertions(+), 19 deletions(-)
diff --git a/tools/flask/policy/policy/flask/access_vectors b/tools/flask/policy/policy/flask/access_vectors
index 644f2e1..5901911 100644
--- a/tools/flask/policy/policy/flask/access_vectors
+++ b/tools/flask/policy/policy/flask/access_vectors
@@ -85,6 +85,7 @@ class domain
getpodtarget
setpodtarget
set_misc_info
+ set_virq_handler
}
class hvm
diff --git a/tools/libxc/xc_domain.c b/tools/libxc/xc_domain.c
index ab019b8..d98e68b 100644
--- a/tools/libxc/xc_domain.c
+++ b/tools/libxc/xc_domain.c
@@ -1504,6 +1504,16 @@ int xc_domain_set_access_required(xc_interface *xch,
return do_domctl(xch, &domctl);
}
+int xc_domain_set_virq_handler(xc_interface *xch, uint32_t domid, int virq)
+{
+ DECLARE_DOMCTL;
+
+ domctl.cmd = XEN_DOMCTL_set_virq_handler;
+ domctl.domain = domid;
+ domctl.u.set_virq_handler.virq = virq;
+ return do_domctl(xch, &domctl);
+}
+
/*
* Local variables:
* mode: C
diff --git a/tools/libxc/xenctrl.h b/tools/libxc/xenctrl.h
index 37b0fd6..2ffdd85 100644
--- a/tools/libxc/xenctrl.h
+++ b/tools/libxc/xenctrl.h
@@ -749,6 +749,15 @@ int xc_domain_p2m_audit(xc_interface *xch,
int xc_domain_set_access_required(xc_interface *xch,
uint32_t domid,
unsigned int required);
+/**
+ * This function sets the handler of global VIRQs sent by the hypervisor
+ *
+ * @parm xch a handle to an open hypervisor interface
+ * @parm domid the domain id which will handle the VIRQ
+ * @parm virq the virq number (VIRQ_*)
+ * return 0 on success, -1 on failure
+ */
+int xc_domain_set_virq_handler(xc_interface *xch, uint32_t domid, int virq);
/*
* CPUPOOL MANAGEMENT FUNCTIONS
diff --git a/xen/arch/x86/cpu/mcheck/amd_nonfatal.c b/xen/arch/x86/cpu/mcheck/amd_nonfatal.c
index 50288bd..9222098 100644
--- a/xen/arch/x86/cpu/mcheck/amd_nonfatal.c
+++ b/xen/arch/x86/cpu/mcheck/amd_nonfatal.c
@@ -100,7 +100,7 @@ static void mce_amd_checkregs(void *info)
if (dom0_vmce_enabled()) {
mctelem_commit(mctc);
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
} else if (++dumpcount >= 10) {
x86_mcinfo_dump((struct mc_info *)mctelem_dataptr(mctc));
mctelem_dismiss(mctc);
diff --git a/xen/arch/x86/cpu/mcheck/mce.c b/xen/arch/x86/cpu/mcheck/mce.c
index b592041..c4e4477 100644
--- a/xen/arch/x86/cpu/mcheck/mce.c
+++ b/xen/arch/x86/cpu/mcheck/mce.c
@@ -594,7 +594,7 @@ void mcheck_cmn_handler(struct cpu_user_regs *regs, long error_code,
if (dom0_vmce_enabled()) {
if (mctc != NULL)
mctelem_commit(mctc);
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
} else {
x86_mcinfo_dump(mci);
if (mctc != NULL)
diff --git a/xen/arch/x86/cpu/mcheck/mce_intel.c b/xen/arch/x86/cpu/mcheck/mce_intel.c
index 0986025..0894080 100644
--- a/xen/arch/x86/cpu/mcheck/mce_intel.c
+++ b/xen/arch/x86/cpu/mcheck/mce_intel.c
@@ -354,7 +354,7 @@ static void mce_softirq(void)
/* Step2: Send Log to DOM0 through vIRQ */
if (dom0_vmce_enabled()) {
mce_printk(MCE_VERBOSE, "MCE: send MCE# to DOM0 through virq\n");
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
}
}
@@ -1085,7 +1085,7 @@ static void cmci_discover(void)
if (bs.errcnt && mctc != NULL) {
if (dom0_vmce_enabled()) {
mctelem_commit(mctc);
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
} else {
x86_mcinfo_dump(mctelem_dataptr(mctc));
mctelem_dismiss(mctc);
@@ -1205,7 +1205,7 @@ fastcall void smp_cmci_interrupt(struct cpu_user_regs *regs)
if (dom0_vmce_enabled()) {
mctelem_commit(mctc);
mce_printk(MCE_VERBOSE, "CMCI: send CMCI to DOM0 through virq\n");
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
} else {
x86_mcinfo_dump(mctelem_dataptr(mctc));
mctelem_dismiss(mctc);
diff --git a/xen/arch/x86/cpu/mcheck/non-fatal.c b/xen/arch/x86/cpu/mcheck/non-fatal.c
index c57688f..1dded9b 100644
--- a/xen/arch/x86/cpu/mcheck/non-fatal.c
+++ b/xen/arch/x86/cpu/mcheck/non-fatal.c
@@ -55,7 +55,7 @@ static void mce_checkregs (void *info)
if (dom0_vmce_enabled()) {
mctelem_commit(mctc);
- send_guest_global_virq(dom0, VIRQ_MCA);
+ send_global_virq(VIRQ_MCA);
} else if (++dumpcount >= 10) {
x86_mcinfo_dump((struct mc_info *)mctelem_dataptr(mctc));
mctelem_dismiss(mctc);
diff --git a/xen/common/cpu.c b/xen/common/cpu.c
index 79abdb7..630881e 100644
--- a/xen/common/cpu.c
+++ b/xen/common/cpu.c
@@ -108,7 +108,7 @@ int cpu_down(unsigned int cpu)
notifier_rc = notifier_call_chain(&cpu_chain, CPU_DEAD, hcpu, NULL);
BUG_ON(notifier_rc != NOTIFY_DONE);
- send_guest_global_virq(dom0, VIRQ_PCPU_STATE);
+ send_global_virq(VIRQ_PCPU_STATE);
cpu_hotplug_done();
return 0;
@@ -148,7 +148,7 @@ int cpu_up(unsigned int cpu)
notifier_rc = notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu, NULL);
BUG_ON(notifier_rc != NOTIFY_DONE);
- send_guest_global_virq(dom0, VIRQ_PCPU_STATE);
+ send_global_virq(VIRQ_PCPU_STATE);
cpu_hotplug_done();
return 0;
diff --git a/xen/common/domain.c b/xen/common/domain.c
index fd20210..500c7a2 100644
--- a/xen/common/domain.c
+++ b/xen/common/domain.c
@@ -86,7 +86,7 @@ static void __domain_finalise_shutdown(struct domain *d)
if ( (d->shutdown_code == SHUTDOWN_suspend) && d->suspend_evtchn )
evtchn_send(d, d->suspend_evtchn);
else
- send_guest_global_virq(dom0, VIRQ_DOM_EXC);
+ send_global_virq(VIRQ_DOM_EXC);
}
static void vcpu_check_shutdown(struct vcpu *v)
@@ -480,7 +480,7 @@ int domain_kill(struct domain *d)
}
d->is_dying = DOMDYING_dead;
put_domain(d);
- send_guest_global_virq(dom0, VIRQ_DOM_EXC);
+ send_global_virq(VIRQ_DOM_EXC);
/* fallthrough */
case DOMDYING_dead:
break;
@@ -621,7 +621,7 @@ void domain_pause_for_debugger(void)
for_each_vcpu ( d, v )
vcpu_sleep_nosync(v);
- send_guest_global_virq(dom0, VIRQ_DEBUGGER);
+ send_global_virq(VIRQ_DEBUGGER);
}
/* Complete domain destroy after RCU readers are not holding old references. */
@@ -680,7 +680,7 @@ static void complete_domain_destroy(struct rcu_head *head)
free_cpumask_var(d->domain_dirty_cpumask);
free_domain_struct(d);
- send_guest_global_virq(dom0, VIRQ_DOM_EXC);
+ send_global_virq(VIRQ_DOM_EXC);
}
/* Release resources belonging to task @p. */
diff --git a/xen/common/domctl.c b/xen/common/domctl.c
index 5b0fc4a..8001a91 100644
--- a/xen/common/domctl.c
+++ b/xen/common/domctl.c
@@ -995,6 +995,23 @@ long do_domctl(XEN_GUEST_HANDLE(xen_domctl_t) u_domctl)
}
break;
+ case XEN_DOMCTL_set_virq_handler:
+ {
+ struct domain *d;
+ uint32_t virq = op->u.set_virq_handler.virq;
+
+ ret = -ESRCH;
+ d = rcu_lock_domain_by_id(op->domain);
+ if ( d != NULL )
+ {
+ ret = xsm_set_virq_handler(d, virq);
+ if ( !ret )
+ ret = set_global_virq_handler(d, virq);
+ rcu_unlock_domain(d);
+ }
+ }
+ break;
+
default:
ret = arch_do_domctl(op, u_domctl);
break;
diff --git a/xen/common/event_channel.c b/xen/common/event_channel.c
index 9212042..43bd167 100644
--- a/xen/common/event_channel.c
+++ b/xen/common/event_channel.c
@@ -689,7 +689,7 @@ void send_guest_vcpu_virq(struct vcpu *v, int virq)
spin_unlock_irqrestore(&v->virq_lock, flags);
}
-void send_guest_global_virq(struct domain *d, int virq)
+static void send_guest_global_virq(struct domain *d, int virq)
{
unsigned long flags;
int port;
@@ -739,6 +739,68 @@ int send_guest_pirq(struct domain *d, const struct pirq *pirq)
return evtchn_set_pending(d->vcpu[chn->notify_vcpu_id], port);
}
+static struct domain *global_virq_handlers[NR_VIRQS] __read_mostly;
+
+static DEFINE_SPINLOCK(global_virq_handlers_lock);
+
+void send_global_virq(uint32_t virq)
+{
+ ASSERT(virq < NR_VIRQS);
+ ASSERT(virq_is_global(virq));
+
+ send_guest_global_virq(global_virq_handlers[virq] ?: dom0, virq);
+}
+
+int set_global_virq_handler(struct domain *d, uint32_t virq)
+{
+ struct domain *old;
+
+ if (virq >= NR_VIRQS)
+ return -EINVAL;
+ if (!virq_is_global(virq))
+ return -EINVAL;
+
+ if (global_virq_handlers[virq] == d)
+ return 0;
+
+ if (unlikely(!get_domain(d)))
+ return -EINVAL;
+
+ spin_lock(&global_virq_handlers_lock);
+ old = global_virq_handlers[virq];
+ global_virq_handlers[virq] = d;
+ spin_unlock(&global_virq_handlers_lock);
+
+ if (old != NULL)
+ put_domain(old);
+
+ return 0;
+}
+
+static void clear_global_virq_handlers(struct domain *d)
+{
+ uint32_t virq;
+ int put_count = 0;
+
+ spin_lock(&global_virq_handlers_lock);
+
+ for (virq = 0; virq < NR_VIRQS; virq++)
+ {
+ if (global_virq_handlers[virq] == d)
+ {
+ global_virq_handlers[virq] = NULL;
+ put_count++;
+ }
+ }
+
+ spin_unlock(&global_virq_handlers_lock);
+
+ while (put_count)
+ {
+ put_domain(d);
+ put_count--;
+ }
+}
static long evtchn_status(evtchn_status_t *status)
{
@@ -1160,6 +1222,8 @@ void evtchn_destroy(struct domain *d)
d->evtchn[i] = NULL;
}
spin_unlock(&d->event_lock);
+
+ clear_global_virq_handlers(d);
}
diff --git a/xen/common/trace.c b/xen/common/trace.c
index 5772f24..58cbf39 100644
--- a/xen/common/trace.c
+++ b/xen/common/trace.c
@@ -661,7 +661,7 @@ static inline void insert_lost_records(struct t_buf *buf)
*/
static void trace_notify_dom0(unsigned long unused)
{
- send_guest_global_virq(dom0, VIRQ_TBUF);
+ send_global_virq(VIRQ_TBUF);
}
static DECLARE_SOFTIRQ_TASKLET(trace_notify_dom0_tasklet,
trace_notify_dom0, 0);
diff --git a/xen/drivers/char/console.c b/xen/drivers/char/console.c
index 89cf4f8..6560182 100644
--- a/xen/drivers/char/console.c
+++ b/xen/drivers/char/console.c
@@ -288,7 +288,7 @@ static void __serial_rx(char c, struct cpu_user_regs *regs)
if ( (serial_rx_prod-serial_rx_cons) != SERIAL_RX_SIZE )
serial_rx_ring[SERIAL_RX_MASK(serial_rx_prod++)] = c;
/* Always notify the guest: prevents receive path from getting stuck. */
- send_guest_global_virq(dom0, VIRQ_CONSOLE);
+ send_global_virq(VIRQ_CONSOLE);
}
static void serial_rx(char c, struct cpu_user_regs *regs)
@@ -315,7 +315,7 @@ static void serial_rx(char c, struct cpu_user_regs *regs)
static void notify_dom0_con_ring(unsigned long unused)
{
- send_guest_global_virq(dom0, VIRQ_CON_RING);
+ send_global_virq(VIRQ_CON_RING);
}
static DECLARE_SOFTIRQ_TASKLET(notify_dom0_con_ring_tasklet,
notify_dom0_con_ring, 0);
diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index c7640aa..75be370 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -813,6 +813,12 @@ struct xen_domctl_audit_p2m {
typedef struct xen_domctl_audit_p2m xen_domctl_audit_p2m_t;
DEFINE_XEN_GUEST_HANDLE(xen_domctl_audit_p2m_t);
+struct xen_domctl_set_virq_handler {
+ uint32_t virq; /* IN */
+};
+typedef struct xen_domctl_set_virq_handler xen_domctl_set_virq_handler_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_virq_handler_t);
+
#if defined(__i386__) || defined(__x86_64__)
/* XEN_DOMCTL_setvcpuextstate */
/* XEN_DOMCTL_getvcpuextstate */
@@ -912,6 +918,7 @@ struct xen_domctl {
#define XEN_DOMCTL_getvcpuextstate 63
#define XEN_DOMCTL_set_access_required 64
#define XEN_DOMCTL_audit_p2m 65
+#define XEN_DOMCTL_set_virq_handler 66
#define XEN_DOMCTL_gdbsx_guestmemio 1000
#define XEN_DOMCTL_gdbsx_pausevcpu 1001
#define XEN_DOMCTL_gdbsx_unpausevcpu 1002
@@ -966,6 +973,7 @@ struct xen_domctl {
#endif
struct xen_domctl_set_access_required access_required;
struct xen_domctl_audit_p2m audit_p2m;
+ struct xen_domctl_set_virq_handler set_virq_handler;
struct xen_domctl_gdbsx_memio gdbsx_guest_memio;
struct xen_domctl_gdbsx_pauseunp_vcpu gdbsx_pauseunp_vcpu;
struct xen_domctl_gdbsx_domstatus gdbsx_domstatus;
diff --git a/xen/include/xen/event.h b/xen/include/xen/event.h
index 232d50e..40b8a7a 100644
--- a/xen/include/xen/event.h
+++ b/xen/include/xen/event.h
@@ -23,11 +23,17 @@
void send_guest_vcpu_virq(struct vcpu *v, int virq);
/*
- * send_guest_global_virq: Notify guest via a global VIRQ.
- * @d: Domain to which virtual IRQ should be sent
+ * send_global_virq: Notify the domain handling a global VIRQ.
* @virq: Virtual IRQ number (VIRQ_*)
*/
-void send_guest_global_virq(struct domain *d, int virq);
+void send_global_virq(uint32_t virq);
+
+/*
+ * sent_global_virq_handler: Set a global VIRQ handler.
+ * @d: New target domain for this VIRQ
+ * @virq: Virtual IRQ number (VIRQ_*), must be global
+ */
+int set_global_virq_handler(struct domain *d, uint32_t virq);
/*
* send_guest_pirq:
diff --git a/xen/include/xsm/xsm.h b/xen/include/xsm/xsm.h
index 566c808..e3cae60 100644
--- a/xen/include/xsm/xsm.h
+++ b/xen/include/xsm/xsm.h
@@ -64,6 +64,7 @@ struct xsm_operations {
int (*domain_settime) (struct domain *d);
int (*set_target) (struct domain *d, struct domain *e);
int (*domctl) (struct domain *d, int cmd);
+ int (*set_virq_handler) (struct domain *d, uint32_t virq);
int (*tbufcontrol) (void);
int (*readconsole) (uint32_t clear);
int (*sched_id) (void);
@@ -265,6 +266,11 @@ static inline int xsm_domctl (struct domain *d, int cmd)
return xsm_call(domctl(d, cmd));
}
+static inline int xsm_set_virq_handler (struct domain *d, uint32_t virq)
+{
+ return xsm_call(set_virq_handler(d, virq));
+}
+
static inline int xsm_tbufcontrol (void)
{
return xsm_call(tbufcontrol());
diff --git a/xen/xsm/dummy.c b/xen/xsm/dummy.c
index 65daa4e..acf9c8a 100644
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -94,6 +94,11 @@ static int dummy_domctl(struct domain *d, int cmd)
return 0;
}
+static int dummy_set_virq_handler(struct domain *d, uint32_t virq)
+{
+ return 0;
+}
+
static int dummy_tbufcontrol (void)
{
return 0;
@@ -596,6 +601,7 @@ void xsm_fixup_ops (struct xsm_operations *ops)
set_to_dummy_if_null(ops, domain_settime);
set_to_dummy_if_null(ops, set_target);
set_to_dummy_if_null(ops, domctl);
+ set_to_dummy_if_null(ops, set_virq_handler);
set_to_dummy_if_null(ops, tbufcontrol);
set_to_dummy_if_null(ops, readconsole);
set_to_dummy_if_null(ops, sched_id);
diff --git a/xen/xsm/flask/hooks.c b/xen/xsm/flask/hooks.c
index a2020a9..543dc77 100644
--- a/xen/xsm/flask/hooks.c
+++ b/xen/xsm/flask/hooks.c
@@ -597,6 +597,11 @@ static int flask_domctl(struct domain *d, int cmd)
return domain_has_perm(current->domain, d, SECCLASS_DOMAIN, DOMAIN__SET_MISC_INFO);
}
+static int flask_set_virq_handler(struct domain *d, uint32_t virq)
+{
+ return domain_has_perm(current->domain, d, SECCLASS_DOMAIN, DOMAIN__SET_VIRQ_HANDLER);
+}
+
static int flask_tbufcontrol(void)
{
return domain_has_xen(current->domain, XEN__TBUFCONTROL);
@@ -1460,6 +1465,7 @@ static struct xsm_operations flask_ops = {
.domain_settime = flask_domain_settime,
.set_target = flask_set_target,
.domctl = flask_domctl,
+ .set_virq_handler = flask_set_virq_handler,
.tbufcontrol = flask_tbufcontrol,
.readconsole = flask_readconsole,
.sched_id = flask_sched_id,
diff --git a/xen/xsm/flask/include/av_perm_to_string.h b/xen/xsm/flask/include/av_perm_to_string.h
index 85cbffc..17a1c36 100644
--- a/xen/xsm/flask/include/av_perm_to_string.h
+++ b/xen/xsm/flask/include/av_perm_to_string.h
@@ -60,6 +60,7 @@
S_(SECCLASS_DOMAIN, DOMAIN__GETPODTARGET, "getpodtarget")
S_(SECCLASS_DOMAIN, DOMAIN__SETPODTARGET, "setpodtarget")
S_(SECCLASS_DOMAIN, DOMAIN__SET_MISC_INFO, "set_misc_info")
+ S_(SECCLASS_DOMAIN, DOMAIN__SET_VIRQ_HANDLER, "set_virq_handler")
S_(SECCLASS_HVM, HVM__SETHVMC, "sethvmc")
S_(SECCLASS_HVM, HVM__GETHVMC, "gethvmc")
S_(SECCLASS_HVM, HVM__SETPARAM, "setparam")
diff --git a/xen/xsm/flask/include/av_permissions.h b/xen/xsm/flask/include/av_permissions.h
index 9e55a86..42eaf81 100644
--- a/xen/xsm/flask/include/av_permissions.h
+++ b/xen/xsm/flask/include/av_permissions.h
@@ -61,6 +61,7 @@
#define DOMAIN__GETPODTARGET 0x10000000UL
#define DOMAIN__SETPODTARGET 0x20000000UL
#define DOMAIN__SET_MISC_INFO 0x40000000UL
+#define DOMAIN__SET_VIRQ_HANDLER 0x80000000UL
#define HVM__SETHVMC 0x00000001UL
#define HVM__GETHVMC 0x00000002UL
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 03/23] xen: change virq parameters from int to uint32_t
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
2012-01-25 14:37 ` [PATCH 01/23] xen: reinstate previously unused XENMEM_remove_from_physmap hypercall Daniel De Graaf
2012-01-25 14:37 ` [PATCH 02/23] xen: allow global VIRQ handlers to be delegated to other domains Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 04/23] xen: use XSM instead of IS_PRIV for getdomaininfo Daniel De Graaf
` (19 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
xen/common/event_channel.c | 10 +++++-----
xen/include/asm-ia64/event.h | 2 +-
xen/include/asm-x86/event.h | 2 +-
xen/include/xen/event.h | 4 ++--
4 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/xen/common/event_channel.c b/xen/common/event_channel.c
index 43bd167..f784254 100644
--- a/xen/common/event_channel.c
+++ b/xen/common/event_channel.c
@@ -104,11 +104,11 @@ static uint8_t get_xen_consumer(xen_event_channel_notification_t fn)
static int evtchn_set_pending(struct vcpu *v, int port);
-static int virq_is_global(int virq)
+static int virq_is_global(uint32_t virq)
{
int rc;
- ASSERT((virq >= 0) && (virq < NR_VIRQS));
+ ASSERT(virq < NR_VIRQS);
switch ( virq )
{
@@ -665,12 +665,12 @@ static int evtchn_set_pending(struct vcpu *v, int port)
return 0;
}
-int guest_enabled_event(struct vcpu *v, int virq)
+int guest_enabled_event(struct vcpu *v, uint32_t virq)
{
return ((v != NULL) && (v->virq_to_evtchn[virq] != 0));
}
-void send_guest_vcpu_virq(struct vcpu *v, int virq)
+void send_guest_vcpu_virq(struct vcpu *v, uint32_t virq)
{
unsigned long flags;
int port;
@@ -689,7 +689,7 @@ void send_guest_vcpu_virq(struct vcpu *v, int virq)
spin_unlock_irqrestore(&v->virq_lock, flags);
}
-static void send_guest_global_virq(struct domain *d, int virq)
+static void send_guest_global_virq(struct domain *d, uint32_t virq)
{
unsigned long flags;
int port;
diff --git a/xen/include/asm-ia64/event.h b/xen/include/asm-ia64/event.h
index c99babd..4463cb3 100644
--- a/xen/include/asm-ia64/event.h
+++ b/xen/include/asm-ia64/event.h
@@ -63,7 +63,7 @@ static inline void local_event_delivery_enable(void)
current->vcpu_info->evtchn_upcall_mask = 0;
}
-static inline int arch_virq_is_global(int virq)
+static inline int arch_virq_is_global(uint32_t virq)
{
int rc;
diff --git a/xen/include/asm-x86/event.h b/xen/include/asm-x86/event.h
index 606ec6d..06057c7 100644
--- a/xen/include/asm-x86/event.h
+++ b/xen/include/asm-x86/event.h
@@ -39,7 +39,7 @@ static inline void local_event_delivery_enable(void)
}
/* No arch specific virq definition now. Default to global. */
-static inline int arch_virq_is_global(int virq)
+static inline int arch_virq_is_global(uint32_t virq)
{
return 1;
}
diff --git a/xen/include/xen/event.h b/xen/include/xen/event.h
index 40b8a7a..22fc6a3 100644
--- a/xen/include/xen/event.h
+++ b/xen/include/xen/event.h
@@ -20,7 +20,7 @@
* @v: VCPU to which virtual IRQ should be sent
* @virq: Virtual IRQ number (VIRQ_*)
*/
-void send_guest_vcpu_virq(struct vcpu *v, int virq);
+void send_guest_vcpu_virq(struct vcpu *v, uint32_t virq);
/*
* send_global_virq: Notify the domain handling a global VIRQ.
@@ -65,7 +65,7 @@ void free_xen_event_channel(
struct vcpu *local_vcpu, int port);
/* Query if event channel is in use by the guest */
-int guest_enabled_event(struct vcpu *v, int virq);
+int guest_enabled_event(struct vcpu *v, uint32_t virq);
/* Notify remote end of a Xen-attached event channel.*/
void notify_via_xen_event_channel(struct domain *ld, int lport);
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 04/23] xen: use XSM instead of IS_PRIV for getdomaininfo
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (2 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 03/23] xen: change virq parameters from int to uint32_t Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 05/23] xen: Preserve reserved grant entries when switching versions Daniel De Graaf
` (18 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
The XEN_DOMCTL_getdomaininfo domctl does not allow manipulation of
domains, only basic information such as size and state, so its use
does not fully justify making a domain privileged. XSM modules can also
provide fine-grained control over what domains are visible to domains
that call getdomaininfo.
If XSM is disabled (either at compile time or by using the dummy XSM
module) then there is no change in behavior: only IS_PRIV domains can
use this domctl. If enabled, the XSM module controls access.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
xen/common/domctl.c | 4 ++++
xen/xsm/dummy.c | 2 ++
2 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/xen/common/domctl.c b/xen/common/domctl.c
index 8001a91..904fb45 100644
--- a/xen/common/domctl.c
+++ b/xen/common/domctl.c
@@ -264,6 +264,10 @@ long do_domctl(XEN_GUEST_HANDLE(xen_domctl_t) u_domctl)
return -EPERM;
break;
}
+#ifdef XSM_ENABLE
+ case XEN_DOMCTL_getdomaininfo:
+ break;
+#endif
default:
if ( !IS_PRIV(current->domain) )
return -EPERM;
diff --git a/xen/xsm/dummy.c b/xen/xsm/dummy.c
index acf9c8a..d99f886 100644
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -66,6 +66,8 @@ static int dummy_scheduler (struct domain *d)
static int dummy_getdomaininfo (struct domain *d)
{
+ if ( !IS_PRIV(current->domain) )
+ return -EPERM;
return 0;
}
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 05/23] xen: Preserve reserved grant entries when switching versions
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (3 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 04/23] xen: use XSM instead of IS_PRIV for getdomaininfo Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 06/23] tools/libxl: pull xenstore/console domids from xenstore Daniel De Graaf
` (17 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
In order for the toolstack to use reserved grant table entries, the
grant table for a guest must be initialized prior to the guest's boot.
When the guest switches grant table versions (necessary if the guest is
using v2 grant tables, or on kexec if switching grant versions), these
initial grants will be cleared. Instead of clearing them, preserve
the grants across the type change.
Attempting to preserve v2-only features such as sub-page grants will
produce a warning and invalidate the resulting v1 grant entry.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
xen/common/grant_table.c | 48 +++++++++++++++++++++++++++++++++----
xen/include/public/grant_table.h | 7 +++++
2 files changed, 49 insertions(+), 6 deletions(-)
diff --git a/xen/common/grant_table.c b/xen/common/grant_table.c
index 0c55fd1..6f24a94 100644
--- a/xen/common/grant_table.c
+++ b/xen/common/grant_table.c
@@ -2111,6 +2111,7 @@ gnttab_set_version(XEN_GUEST_HANDLE(gnttab_set_version_t uop))
struct domain *d = current->domain;
struct grant_table *gt = d->grant_table;
struct active_grant_entry *act;
+ grant_entry_v1_t reserved_entries[GNTTAB_NR_RESERVED_ENTRIES];
long res;
int i;
@@ -2131,7 +2132,7 @@ gnttab_set_version(XEN_GUEST_HANDLE(gnttab_set_version_t uop))
/* (You need to change the version number for e.g. kexec.) */
if ( gt->gt_version != 0 )
{
- for ( i = 0; i < nr_grant_entries(gt); i++ )
+ for ( i = GNTTAB_NR_RESERVED_ENTRIES; i < nr_grant_entries(gt); i++ )
{
act = &active_entry(gt, i);
if ( act->pin != 0 )
@@ -2156,15 +2157,50 @@ gnttab_set_version(XEN_GUEST_HANDLE(gnttab_set_version_t uop))
goto out_unlock;
}
+ /* Preserve the first 8 entries (toolstack reserved grants) */
+ if (gt->gt_version == 1)
+ {
+ memcpy(reserved_entries, gt->shared_v1[0], sizeof(reserved_entries));
+ }
+ else if (gt->gt_version == 2)
+ {
+ for ( i = 0; i < GNTTAB_NR_RESERVED_ENTRIES && i < nr_grant_entries(gt); i++ )
+ {
+ reserved_entries[i].flags = shared_entry_v2(gt, i).hdr.flags;
+ reserved_entries[i].domid = shared_entry_v2(gt, i).hdr.domid;
+ reserved_entries[i].frame = shared_entry_v2(gt, i).full_page.frame;
+ reserved_entries[i].flags |= status_entry(gt, i);
+ if ((reserved_entries[i].flags & GTF_type_mask) > GTF_permit_access)
+ {
+ gdprintk(XENLOG_INFO, "d%d: bad flags %x in grant %d when switching grant version\n",
+ d->domain_id, reserved_entries[i].flags, i);
+ reserved_entries[i].flags = GTF_invalid;
+ }
+ }
+ }
+
if ( op.version < 2 && gt->gt_version == 2 )
gnttab_unpopulate_status_frames(d, gt);
- if ( op.version != gt->gt_version )
+ /* Make sure there's no crud left over in the table from the
+ old version. */
+ for ( i = 0; i < nr_grant_frames(gt); i++ )
+ memset(gt->shared_raw[i], 0, PAGE_SIZE);
+
+ /* Restore the first 8 entries (toolstack reserved grants) */
+ if (gt->gt_version != 0 && op.version == 1)
{
- /* Make sure there's no crud left over in the table from the
- old version. */
- for ( i = 0; i < nr_grant_frames(gt); i++ )
- memset(gt->shared_raw[i], 0, PAGE_SIZE);
+ memcpy(gt->shared_v1[0], reserved_entries, sizeof(reserved_entries));
+ }
+ else if (gt->gt_version != 0 && op.version == 2)
+ {
+ for ( i = 0; i < GNTTAB_NR_RESERVED_ENTRIES; i++ )
+ {
+ status_entry(gt, i) = reserved_entries[i].flags & (GTF_reading|GTF_writing);
+ shared_entry_v2(gt, i).hdr.flags = reserved_entries[i].flags & ~(GTF_reading|GTF_writing);
+ shared_entry_v2(gt, i).hdr.domid = reserved_entries[i].domid;
+ shared_entry_v2(gt, i).full_page.frame = reserved_entries[i].frame;
+ }
}
gt->gt_version = op.version;
diff --git a/xen/include/public/grant_table.h b/xen/include/public/grant_table.h
index 54d9551..292d724 100644
--- a/xen/include/public/grant_table.h
+++ b/xen/include/public/grant_table.h
@@ -117,6 +117,13 @@ struct grant_entry_v1 {
};
typedef struct grant_entry_v1 grant_entry_v1_t;
+/* The first few grant table entries will be preserved across grant table
+ * version changes and may be pre-populated at domain creation by tools.
+ */
+#define GNTTAB_NR_RESERVED_ENTRIES 8
+#define GNTTAB_RESERVED_CONSOLE 0
+#define GNTTAB_RESERVED_XENSTORE 1
+
/*
* Type of grant entry.
* GTF_invalid: This grant entry grants no privileges.
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 06/23] tools/libxl: pull xenstore/console domids from xenstore
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (4 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 05/23] xen: Preserve reserved grant entries when switching versions Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 07/23] lib{xc, xl}: Seed grant tables with xenstore and console grants Daniel De Graaf
` (16 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Instead of assuming that xenstored and xenconsoled are running in dom0,
pull the domain IDs from xenstore.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/libxl/libxl_dom.c | 14 ++++++++++++--
tools/libxl/libxl_internal.h | 2 ++
2 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
index 91643a2..3828c62 100644
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -64,6 +64,7 @@ int libxl__build_pre(libxl__gc *gc, uint32_t domid,
{
libxl_ctx *ctx = libxl__gc_owner(gc);
int tsc_mode;
+ char *xs_domid, *con_domid;
xc_domain_max_vcpus(ctx->xch, domid, info->max_vcpus);
xc_domain_setmaxmem(ctx->xch, domid, info->target_memkb + LIBXL_MAXMEM_CONSTANT);
if (info->type == LIBXL_DOMAIN_TYPE_PV)
@@ -95,9 +96,18 @@ int libxl__build_pre(libxl__gc *gc, uint32_t domid,
xc_shadow_control(ctx->xch, domid, XEN_DOMCTL_SHADOW_OP_SET_ALLOCATION, NULL, 0, &shadow, 0, NULL);
}
- state->store_port = xc_evtchn_alloc_unbound(ctx->xch, domid, 0);
- state->console_port = xc_evtchn_alloc_unbound(ctx->xch, domid, 0);
+ xs_domid = xs_read(ctx->xsh, XBT_NULL, "/tool/xenstored/domid", NULL);
+ state->store_domid = xs_domid ? atoi(xs_domid) : 0;
+ free(xs_domid);
+
+ con_domid = xs_read(ctx->xsh, XBT_NULL, "/tool/xenconsoled/domid", NULL);
+ state->console_domid = con_domid ? atoi(con_domid) : 0;
+ free(con_domid);
+
+ state->store_port = xc_evtchn_alloc_unbound(ctx->xch, domid, state->store_domid);
+ state->console_port = xc_evtchn_alloc_unbound(ctx->xch, domid, state->console_domid);
state->vm_generationid_addr = 0;
+
return 0;
}
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
index 3c8da45..965dbd0 100644
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -243,9 +243,11 @@ _hidden int libxl__domain_shutdown_reason(libxl__gc *gc, uint32_t domid);
libxl__domain_type((gc), (domid)) == LIBXL_DOMAIN_TYPE_##type
typedef struct {
uint32_t store_port;
+ uint32_t store_domid;
unsigned long store_mfn;
uint32_t console_port;
+ uint32_t console_domid;
unsigned long console_mfn;
unsigned long vm_generationid_addr;
} libxl__domain_build_state;
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 07/23] lib{xc, xl}: Seed grant tables with xenstore and console grants
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (5 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 06/23] tools/libxl: pull xenstore/console domids from xenstore Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 08/23] mini-os: avoid crash if no console is provided Daniel De Graaf
` (15 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
This patch claims one reserved grant entry for the console and another
for the xenstore. It modifies the builder to fill in the grant table
entries for the console and the xenstore.
Previous versions of this patch have been sent to xen-devel. See
http://lists.xensource.com/archives/html/xen-devel/2008-07/msg00610.html
http://lists.xensource.com/archives/html/xen-devel/2009-03/msg01491.html
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/libxc/xc_dom.h | 13 +++
tools/libxc/xc_dom_boot.c | 158 +++++++++++++++++++++++++++++++++++++
tools/libxc/xc_dom_compat_linux.c | 2 +
tools/libxc/xc_domain_restore.c | 19 ++++-
tools/libxc/xenguest.h | 3 +-
tools/libxl/libxl_dom.c | 18 ++++-
tools/xcutils/xc_restore.c | 4 +-
7 files changed, 210 insertions(+), 7 deletions(-)
diff --git a/tools/libxc/xc_dom.h b/tools/libxc/xc_dom.h
index e72f066..6c36403 100644
--- a/tools/libxc/xc_dom.h
+++ b/tools/libxc/xc_dom.h
@@ -106,7 +106,9 @@ struct xc_dom_image {
/* misc xen domain config stuff */
unsigned long flags;
unsigned int console_evtchn;
+ unsigned int console_domid;
unsigned int xenstore_evtchn;
+ unsigned int xenstore_domid;
xen_pfn_t shared_info_mfn;
xc_interface *xch;
@@ -200,6 +202,17 @@ void *xc_dom_boot_domU_map(struct xc_dom_image *dom, xen_pfn_t pfn,
xen_pfn_t count);
int xc_dom_boot_image(struct xc_dom_image *dom);
int xc_dom_compat_check(struct xc_dom_image *dom);
+int xc_dom_gnttab_init(struct xc_dom_image *dom);
+int xc_dom_gnttab_hvm_seed(xc_interface *xch, uint32_t domid,
+ unsigned long console_gmfn,
+ unsigned long xenstore_gmfn,
+ uint32_t console_domid,
+ uint32_t xenstore_domid);
+int xc_dom_gnttab_seed(xc_interface *xch, uint32_t domid,
+ unsigned long console_gmfn,
+ unsigned long xenstore_gmfn,
+ uint32_t console_domid,
+ uint32_t xenstore_domid);
/* --- debugging bits ---------------------------------------------- */
diff --git a/tools/libxc/xc_dom_boot.c b/tools/libxc/xc_dom_boot.c
index 65f60df..5f609de 100644
--- a/tools/libxc/xc_dom_boot.c
+++ b/tools/libxc/xc_dom_boot.c
@@ -34,6 +34,7 @@
#include "xg_private.h"
#include "xc_dom.h"
#include <xen/hvm/params.h>
+#include <xen/grant_table.h>
/* ------------------------------------------------------------------------ */
@@ -275,6 +276,163 @@ int xc_dom_boot_image(struct xc_dom_image *dom)
return rc;
}
+static unsigned long xc_dom_gnttab_setup(xc_interface *xch, uint32_t domid)
+{
+ gnttab_setup_table_t setup;
+ DECLARE_HYPERCALL_BUFFER(unsigned long, gmfnp);
+ int rc;
+ unsigned long gmfn;
+
+ gmfnp = xc_hypercall_buffer_alloc(xch, gmfnp, sizeof(*gmfnp));
+ if (gmfnp == NULL)
+ return -1;
+
+ setup.dom = domid;
+ setup.nr_frames = 1;
+ set_xen_guest_handle(setup.frame_list, gmfnp);
+ setup.status = 0;
+
+ rc = xc_gnttab_op(xch, GNTTABOP_setup_table, &setup, sizeof(setup), 1);
+ gmfn = *gmfnp;
+ xc_hypercall_buffer_free(xch, gmfnp);
+
+ if ( rc != 0 || setup.status != GNTST_okay )
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to setup domU grant table "
+ "[errno=%d, status=%" PRId16 "]\n",
+ __FUNCTION__, rc != 0 ? errno : 0, setup.status);
+ return -1;
+ }
+
+ return gmfn;
+}
+
+int xc_dom_gnttab_seed(xc_interface *xch, uint32_t domid,
+ unsigned long console_gmfn,
+ unsigned long xenstore_gmfn,
+ uint32_t console_domid,
+ uint32_t xenstore_domid)
+{
+
+ unsigned long gnttab_gmfn;
+ grant_entry_v1_t *gnttab;
+
+ gnttab_gmfn = xc_dom_gnttab_setup(xch, domid);
+ if ( gnttab_gmfn == -1 )
+ return -1;
+
+ gnttab = xc_map_foreign_range(xch,
+ domid,
+ PAGE_SIZE,
+ PROT_READ|PROT_WRITE,
+ gnttab_gmfn);
+ if ( gnttab == NULL )
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to map domU grant table "
+ "[errno=%d]\n",
+ __FUNCTION__, errno);
+ return -1;
+ }
+
+ if ( domid != console_domid && console_gmfn != -1)
+ {
+ gnttab[GNTTAB_RESERVED_CONSOLE].flags = GTF_permit_access;
+ gnttab[GNTTAB_RESERVED_CONSOLE].domid = console_domid;
+ gnttab[GNTTAB_RESERVED_CONSOLE].frame = console_gmfn;
+ }
+ if ( domid != xenstore_domid && xenstore_gmfn != -1)
+ {
+ gnttab[GNTTAB_RESERVED_XENSTORE].flags = GTF_permit_access;
+ gnttab[GNTTAB_RESERVED_XENSTORE].domid = xenstore_domid;
+ gnttab[GNTTAB_RESERVED_XENSTORE].frame = xenstore_gmfn;
+ }
+
+ if ( munmap(gnttab, PAGE_SIZE) == -1 )
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to unmap domU grant table "
+ "[errno=%d]\n",
+ __FUNCTION__, errno);
+ return -1;
+ }
+
+ return 0;
+}
+
+int xc_dom_gnttab_hvm_seed(xc_interface *xch, uint32_t domid,
+ unsigned long console_gpfn,
+ unsigned long xenstore_gpfn,
+ uint32_t console_domid,
+ uint32_t xenstore_domid)
+{
+#define SCRATCH_PFN_GNTTAB 0xFFFFE
+
+ int rc;
+ struct xen_add_to_physmap xatp = {
+ .domid = domid,
+ .space = XENMAPSPACE_grant_table,
+ .idx = 0,
+ .gpfn = SCRATCH_PFN_GNTTAB
+ };
+ struct xen_remove_from_physmap xrfp = {
+ .domid = domid,
+ .gpfn = SCRATCH_PFN_GNTTAB
+ };
+
+ rc = do_memory_op(xch, XENMEM_add_to_physmap, &xatp, sizeof(xatp));
+ if ( rc != 0 )
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to add gnttab to physmap "
+ "[errno=%d]\n",
+ __FUNCTION__, errno);
+ return -1;
+ }
+
+ rc = xc_dom_gnttab_seed(xch, domid,
+ console_gpfn, xenstore_gpfn,
+ console_domid, xenstore_domid);
+ if (rc != 0)
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to seed gnttab entries\n",
+ __FUNCTION__);
+ (void) do_memory_op(xch, XENMEM_remove_from_physmap, &xrfp, sizeof(xrfp));
+ return -1;
+ }
+
+ rc = do_memory_op(xch, XENMEM_remove_from_physmap, &xrfp, sizeof(xrfp));
+ if (rc != 0)
+ {
+ xc_dom_panic(xch, XC_INTERNAL_ERROR,
+ "%s: failed to remove gnttab from physmap "
+ "[errno=%d]\n",
+ __FUNCTION__, errno);
+ return -1;
+ }
+
+ return 0;
+}
+
+int xc_dom_gnttab_init(struct xc_dom_image *dom)
+{
+ unsigned long console_gmfn;
+ unsigned long xenstore_gmfn;
+ int autotranslated;
+
+ autotranslated = xc_dom_feature_translated(dom);
+ console_gmfn = autotranslated ?
+ dom->console_pfn : xc_dom_p2m_host(dom, dom->console_pfn);
+ xenstore_gmfn = autotranslated ?
+ dom->xenstore_pfn : xc_dom_p2m_host(dom, dom->xenstore_pfn);
+
+ return xc_dom_gnttab_seed(dom->xch, dom->guest_domid,
+ console_gmfn, xenstore_gmfn,
+ dom->console_domid, dom->xenstore_domid);
+}
+
/*
* Local variables:
* mode: C
diff --git a/tools/libxc/xc_dom_compat_linux.c b/tools/libxc/xc_dom_compat_linux.c
index 0e78842..2183a3b 100644
--- a/tools/libxc/xc_dom_compat_linux.c
+++ b/tools/libxc/xc_dom_compat_linux.c
@@ -62,6 +62,8 @@ static int xc_linux_build_internal(struct xc_dom_image *dom,
goto out;
if ( (rc = xc_dom_boot_image(dom)) != 0 )
goto out;
+ if ( (rc = xc_dom_gnttab_init(dom)) != 0)
+ goto out;
*console_mfn = xc_dom_p2m_host(dom, dom->console_pfn);
*store_mfn = xc_dom_p2m_host(dom, dom->xenstore_pfn);
diff --git a/tools/libxc/xc_domain_restore.c b/tools/libxc/xc_domain_restore.c
index 3fda6f8..8bee684 100644
--- a/tools/libxc/xc_domain_restore.c
+++ b/tools/libxc/xc_domain_restore.c
@@ -1259,7 +1259,8 @@ static int apply_batch(xc_interface *xch, uint32_t dom, struct restore_ctx *ctx,
int xc_domain_restore(xc_interface *xch, int io_fd, uint32_t dom,
unsigned int store_evtchn, unsigned long *store_mfn,
- unsigned int console_evtchn, unsigned long *console_mfn,
+ uint32_t store_domid, unsigned int console_evtchn,
+ unsigned long *console_mfn, uint32_t console_domid,
unsigned int hvm, unsigned int pae, int superpages,
int no_incr_generationid,
unsigned long *vm_generationid_addr)
@@ -2018,6 +2019,14 @@ int xc_domain_restore(xc_interface *xch, int io_fd, uint32_t dom,
memcpy(ctx->live_p2m, ctx->p2m, dinfo->p2m_size * sizeof(xen_pfn_t));
munmap(ctx->live_p2m, P2M_FL_ENTRIES * PAGE_SIZE);
+ rc = xc_dom_gnttab_seed(xch, dom, *console_mfn, *store_mfn,
+ console_domid, store_domid);
+ if (rc != 0)
+ {
+ ERROR("error seeding grant table");
+ goto out;
+ }
+
DPRINTF("Domain ready to be built.\n");
rc = 0;
goto out;
@@ -2076,6 +2085,14 @@ int xc_domain_restore(xc_interface *xch, int io_fd, uint32_t dom,
goto out;
}
+ rc = xc_dom_gnttab_hvm_seed(xch, dom, *console_mfn, *store_mfn,
+ console_domid, store_domid);
+ if (rc != 0)
+ {
+ ERROR("error seeding grant table");
+ goto out;
+ }
+
/* HVM success! */
rc = 0;
diff --git a/tools/libxc/xenguest.h b/tools/libxc/xenguest.h
index 6026370..3bd5549 100644
--- a/tools/libxc/xenguest.h
+++ b/tools/libxc/xenguest.h
@@ -79,7 +79,8 @@ int xc_domain_save(xc_interface *xch, int io_fd, uint32_t dom, uint32_t max_iter
*/
int xc_domain_restore(xc_interface *xch, int io_fd, uint32_t dom,
unsigned int store_evtchn, unsigned long *store_mfn,
- unsigned int console_evtchn, unsigned long *console_mfn,
+ uint32_t store_domid, unsigned int console_evtchn,
+ unsigned long *console_mfn, uint32_t console_domid,
unsigned int hvm, unsigned int pae, int superpages,
int no_incr_generationid,
unsigned long *vm_generationid_addr);
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
index 3828c62..28646e0 100644
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -224,7 +224,9 @@ int libxl__build_pv(libxl__gc *gc, uint32_t domid,
dom->flags = flags;
dom->console_evtchn = state->console_port;
+ dom->console_domid = state->console_domid;
dom->xenstore_evtchn = state->store_port;
+ dom->xenstore_domid = state->store_domid;
if ( (ret = xc_dom_boot_xen_init(dom, ctx->xch, domid)) != 0 ) {
LIBXL__LOG_ERRNO(ctx, LIBXL__LOG_ERROR, "xc_dom_boot_xen_init failed");
@@ -250,6 +252,10 @@ int libxl__build_pv(libxl__gc *gc, uint32_t domid,
LIBXL__LOG_ERRNO(ctx, LIBXL__LOG_ERROR, "xc_dom_boot_image failed");
goto out;
}
+ if ( (ret = xc_dom_gnttab_init(dom)) != 0 ) {
+ LIBXL__LOG_ERRNO(ctx, LIBXL__LOG_ERROR, "xc_dom_gnttab_init failed");
+ goto out;
+ }
state->console_mfn = xc_dom_p2m_host(dom, dom->console_pfn);
state->store_mfn = xc_dom_p2m_host(dom, dom->xenstore_pfn);
@@ -263,7 +269,8 @@ out:
static int hvm_build_set_params(xc_interface *handle, uint32_t domid,
libxl_domain_build_info *info,
int store_evtchn, unsigned long *store_mfn,
- int console_evtchn, unsigned long *console_mfn)
+ int console_evtchn, unsigned long *console_mfn,
+ uint32_t store_domid, uint32_t console_domid)
{
struct hvm_info_table *va_hvm;
uint8_t *va_map, sum;
@@ -296,6 +303,8 @@ static int hvm_build_set_params(xc_interface *handle, uint32_t domid,
xc_set_hvm_param(handle, domid, HVM_PARAM_NESTEDHVM, info->u.hvm.nested_hvm);
xc_set_hvm_param(handle, domid, HVM_PARAM_STORE_EVTCHN, store_evtchn);
xc_set_hvm_param(handle, domid, HVM_PARAM_CONSOLE_EVTCHN, console_evtchn);
+
+ xc_dom_gnttab_hvm_seed(handle, domid, *console_mfn, *store_mfn, console_domid, store_domid);
return 0;
}
@@ -349,7 +358,9 @@ int libxl__build_hvm(libxl__gc *gc, uint32_t domid,
goto out;
}
ret = hvm_build_set_params(ctx->xch, domid, info, state->store_port,
- &state->store_mfn, state->console_port, &state->console_mfn);
+ &state->store_mfn, state->console_port,
+ &state->console_mfn, state->store_domid,
+ state->console_domid);
if (ret) {
LIBXL__LOG_ERRNOVAL(ctx, LIBXL__LOG_ERROR, ret, "hvm build set params failed");
goto out;
@@ -387,7 +398,8 @@ int libxl__domain_restore_common(libxl__gc *gc, uint32_t domid,
}
rc = xc_domain_restore(ctx->xch, fd, domid,
state->store_port, &state->store_mfn,
- state->console_port, &state->console_mfn,
+ state->store_domid, state->console_port,
+ &state->console_mfn, state->console_domid,
hvm, pae, superpages, no_incr_generationid,
&state->vm_generationid_addr);
if ( rc ) {
diff --git a/tools/xcutils/xc_restore.c b/tools/xcutils/xc_restore.c
index 63d53a8..e41a133 100644
--- a/tools/xcutils/xc_restore.c
+++ b/tools/xcutils/xc_restore.c
@@ -45,8 +45,8 @@ main(int argc, char **argv)
else
superpages = !!hvm;
- ret = xc_domain_restore(xch, io_fd, domid, store_evtchn, &store_mfn,
- console_evtchn, &console_mfn, hvm, pae, superpages,
+ ret = xc_domain_restore(xch, io_fd, domid, store_evtchn, &store_mfn, 0,
+ console_evtchn, &console_mfn, 0, hvm, pae, superpages,
0, NULL);
if ( ret == 0 )
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 08/23] mini-os: avoid crash if no console is provided
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (6 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 07/23] lib{xc, xl}: Seed grant tables with xenstore and console grants Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 09/23] mini-os: remove per-fd evtchn limit Daniel De Graaf
` (14 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
| 7 ++++++-
1 files changed, 6 insertions(+), 1 deletions(-)
--git a/extras/mini-os/console/xencons_ring.c b/extras/mini-os/console/xencons_ring.c
index 22fd618..af0afed 100644
--- a/extras/mini-os/console/xencons_ring.c
+++ b/extras/mini-os/console/xencons_ring.c
@@ -25,7 +25,10 @@ static inline void notify_daemon(struct consfront_dev *dev)
static inline struct xencons_interface *xencons_interface(void)
{
- return mfn_to_virt(start_info.console.domU.mfn);
+ if (start_info.console.domU.evtchn)
+ return mfn_to_virt(start_info.console.domU.mfn);
+ else
+ return NULL;
}
int xencons_ring_send_no_notify(struct consfront_dev *dev, const char *data, unsigned len)
@@ -38,6 +41,8 @@ int xencons_ring_send_no_notify(struct consfront_dev *dev, const char *data, uns
intf = xencons_interface();
else
intf = dev->ring;
+ if (!intf)
+ return sent;
cons = intf->out_cons;
prod = intf->out_prod;
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 09/23] mini-os: remove per-fd evtchn limit
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (7 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 08/23] mini-os: avoid crash if no console is provided Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 10/23] mini-os: create app-specific configuration Daniel De Graaf
` (13 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
This changes the minios evtchn implementation to use a list instead of
an array which ahis allows it to grow as necessary to support any number
of ports, only limited by Xen (NR_EVS is 1024, should be enough for now).
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
| 16 +++---
tools/libxc/xc_minios.c | 139 ++++++++++++++++++++++--------------------
2 files changed, 81 insertions(+), 74 deletions(-)
--git a/extras/mini-os/include/lib.h b/extras/mini-os/include/lib.h
index bd3eeaf..12070c3 100644
--- a/extras/mini-os/include/lib.h
+++ b/extras/mini-os/include/lib.h
@@ -53,6 +53,7 @@
#include <xen/xen.h>
#include <xen/event_channel.h>
#include "gntmap.h"
+#include "list.h"
#ifdef HAVE_LIBC
#include <stdio.h>
@@ -143,7 +144,12 @@ enum fd_type {
FTYPE_SAVEFILE,
};
-#define MAX_EVTCHN_PORTS 16
+struct evtchn_port_info {
+ struct minios_list_head list;
+ evtchn_port_t port;
+ unsigned long pending;
+ int bound;
+};
extern struct file {
enum fd_type type;
@@ -158,13 +164,7 @@ extern struct file {
off_t offset;
} file;
struct {
- /* To each event channel FD is associated a series of ports which
- * wakes select for this FD. */
- struct {
- evtchn_port_t port;
- unsigned long pending;
- int bound;
- } ports[MAX_EVTCHN_PORTS];
+ struct minios_list_head ports;
} evtchn;
struct gntmap gntmap;
struct {
diff --git a/tools/libxc/xc_minios.c b/tools/libxc/xc_minios.c
index 8bbfd18..29cce63 100644
--- a/tools/libxc/xc_minios.c
+++ b/tools/libxc/xc_minios.c
@@ -210,15 +210,34 @@ static struct xc_osdep_ops minios_privcmd_ops = {
},
};
+
+/* XXX Note: This is not threadsafe */
+static struct evtchn_port_info* port_alloc(int fd) {
+ struct evtchn_port_info *port_info;
+ port_info = malloc(sizeof(struct evtchn_port_info));
+ if (port_info == NULL)
+ return NULL;
+ port_info->pending = 0;
+ port_info->port = -1;
+ port_info->bound = 0;
+
+ minios_list_add(&port_info->list, &files[fd].evtchn.ports);
+ return port_info;
+}
+
+static void port_dealloc(struct evtchn_port_info *port_info) {
+ if (port_info->bound)
+ unbind_evtchn(port_info->port);
+ minios_list_del(&port_info->list);
+ free(port_info);
+}
+
static xc_osdep_handle minios_evtchn_open(xc_evtchn *xce)
{
- int fd = alloc_fd(FTYPE_EVTCHN), i;
+ int fd = alloc_fd(FTYPE_EVTCHN);
if ( fd == -1 )
return XC_OSDEP_OPEN_ERROR;
- for (i = 0; i < MAX_EVTCHN_PORTS; i++) {
- files[fd].evtchn.ports[i].port = -1;
- files[fd].evtchn.ports[i].bound = 0;
- }
+ MINIOS_INIT_LIST_HEAD(&files[fd].evtchn.ports);
printf("evtchn_open() -> %d\n", fd);
return (xc_osdep_handle)fd;
}
@@ -231,10 +250,10 @@ static int minios_evtchn_close(xc_evtchn *xce, xc_osdep_handle h)
void minios_evtchn_close_fd(int fd)
{
- int i;
- for (i = 0; i < MAX_EVTCHN_PORTS; i++)
- if (files[fd].evtchn.ports[i].bound)
- unbind_evtchn(files[fd].evtchn.ports[i].port);
+ struct evtchn_port_info *port_info, *tmp;
+ minios_list_for_each_entry_safe(port_info, tmp, &files[fd].evtchn.ports, list)
+ port_dealloc(port_info);
+
files[fd].type = FTYPE_NONE;
}
@@ -256,35 +275,21 @@ static int minios_evtchn_notify(xc_evtchn *xce, xc_osdep_handle h, evtchn_port_t
return ret;
}
-/* XXX Note: This is not threadsafe */
-static int port_alloc(int fd) {
- int i;
- for (i= 0; i < MAX_EVTCHN_PORTS; i++)
- if (files[fd].evtchn.ports[i].port == -1)
- break;
- if (i == MAX_EVTCHN_PORTS) {
- printf("Too many ports in xc handle\n");
- errno = EMFILE;
- return -1;
- }
- files[fd].evtchn.ports[i].pending = 0;
- return i;
-}
-
static void evtchn_handler(evtchn_port_t port, struct pt_regs *regs, void *data)
{
int fd = (int)(intptr_t)data;
- int i;
+ struct evtchn_port_info *port_info;
assert(files[fd].type == FTYPE_EVTCHN);
mask_evtchn(port);
- for (i= 0; i < MAX_EVTCHN_PORTS; i++)
- if (files[fd].evtchn.ports[i].port == port)
- break;
- if (i == MAX_EVTCHN_PORTS) {
- printk("Unknown port for handle %d\n", fd);
- return;
+ minios_list_for_each_entry(port_info, &files[fd].evtchn.ports, list) {
+ if (port_info->port == port)
+ goto found;
}
- files[fd].evtchn.ports[i].pending = 1;
+ printk("Unknown port for handle %d\n", fd);
+ return;
+
+ found:
+ port_info->pending = 1;
files[fd].read = 1;
wake_up(&event_queue);
}
@@ -292,12 +297,13 @@ static void evtchn_handler(evtchn_port_t port, struct pt_regs *regs, void *data)
static evtchn_port_or_error_t minios_evtchn_bind_unbound_port(xc_evtchn *xce, xc_osdep_handle h, int domid)
{
int fd = (int)h;
- int ret, i;
+ struct evtchn_port_info *port_info;
+ int ret;
evtchn_port_t port;
assert(get_current() == main_thread);
- i = port_alloc(fd);
- if (i == -1)
+ port_info = port_alloc(fd);
+ if (port_info == NULL)
return -1;
printf("xc_evtchn_bind_unbound_port(%d)", domid);
@@ -305,11 +311,12 @@ static evtchn_port_or_error_t minios_evtchn_bind_unbound_port(xc_evtchn *xce, xc
printf(" = %d\n", ret);
if (ret < 0) {
+ port_dealloc(port_info);
errno = -ret;
return -1;
}
- files[fd].evtchn.ports[i].bound = 1;
- files[fd].evtchn.ports[i].port = port;
+ port_info->bound = 1;
+ port_info->port = port;
unmask_evtchn(port);
return port;
}
@@ -318,12 +325,13 @@ static evtchn_port_or_error_t minios_evtchn_bind_interdomain(xc_evtchn *xce, xc_
evtchn_port_t remote_port)
{
int fd = (int)h;
+ struct evtchn_port_info *port_info;
evtchn_port_t local_port;
- int ret, i;
+ int ret;
assert(get_current() == main_thread);
- i = port_alloc(fd);
- if (i == -1)
+ port_info = port_alloc(fd);
+ if (port_info == NULL)
return -1;
printf("xc_evtchn_bind_interdomain(%d, %"PRId32")", domid, remote_port);
@@ -331,11 +339,12 @@ static evtchn_port_or_error_t minios_evtchn_bind_interdomain(xc_evtchn *xce, xc_
printf(" = %d\n", ret);
if (ret < 0) {
+ port_dealloc(port_info);
errno = -ret;
return -1;
}
- files[fd].evtchn.ports[i].bound = 1;
- files[fd].evtchn.ports[i].port = local_port;
+ port_info->bound = 1;
+ port_info->port = local_port;
unmask_evtchn(local_port);
return local_port;
}
@@ -343,42 +352,40 @@ static evtchn_port_or_error_t minios_evtchn_bind_interdomain(xc_evtchn *xce, xc_
static int minios_evtchn_unbind(xc_evtchn *xce, xc_osdep_handle h, evtchn_port_t port)
{
int fd = (int)h;
- int i;
- for (i = 0; i < MAX_EVTCHN_PORTS; i++)
- if (files[fd].evtchn.ports[i].port == port) {
- files[fd].evtchn.ports[i].port = -1;
- break;
- }
- if (i == MAX_EVTCHN_PORTS) {
- printf("Warning: couldn't find port %"PRId32" for xc handle %x\n", port, fd);
- errno = -EINVAL;
- return -1;
+ struct evtchn_port_info *port_info;
+
+ minios_list_for_each_entry(port_info, &files[fd].evtchn.ports, list) {
+ if (port_info->port == port) {
+ port_dealloc(port_info);
+ return 0;
+ }
}
- files[fd].evtchn.ports[i].bound = 0;
- unbind_evtchn(port);
- return 0;
+ printf("Warning: couldn't find port %"PRId32" for xc handle %x\n", port, fd);
+ errno = -EINVAL;
+ return -1;
}
static evtchn_port_or_error_t minios_evtchn_bind_virq(xc_evtchn *xce, xc_osdep_handle h, unsigned int virq)
{
int fd = (int)h;
+ struct evtchn_port_info *port_info;
evtchn_port_t port;
- int i;
assert(get_current() == main_thread);
- i = port_alloc(fd);
- if (i == -1)
+ port_info = port_alloc(fd);
+ if (port_info == NULL)
return -1;
printf("xc_evtchn_bind_virq(%d)", virq);
port = bind_virq(virq, evtchn_handler, (void*)(intptr_t)fd);
if (port < 0) {
+ port_dealloc(port_info);
errno = -port;
return -1;
}
- files[fd].evtchn.ports[i].bound = 1;
- files[fd].evtchn.ports[i].port = port;
+ port_info->bound = 1;
+ port_info->port = port;
unmask_evtchn(port);
return port;
}
@@ -386,18 +393,18 @@ static evtchn_port_or_error_t minios_evtchn_bind_virq(xc_evtchn *xce, xc_osdep_h
static evtchn_port_or_error_t minios_evtchn_pending(xc_evtchn *xce, xc_osdep_handle h)
{
int fd = (int)h;
- int i;
+ struct evtchn_port_info *port_info;
unsigned long flags;
evtchn_port_t ret = -1;
local_irq_save(flags);
files[fd].read = 0;
- for (i = 0; i < MAX_EVTCHN_PORTS; i++) {
- evtchn_port_t port = files[fd].evtchn.ports[i].port;
- if (port != -1 && files[fd].evtchn.ports[i].pending) {
+
+ minios_list_for_each_entry(port_info, &files[fd].evtchn.ports, list) {
+ if (port_info->port != -1 && port_info->pending) {
if (ret == -1) {
- ret = port;
- files[fd].evtchn.ports[i].pending = 0;
+ ret = port_info->port;
+ port_info->pending = 0;
} else {
files[fd].read = 1;
break;
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 10/23] mini-os: create app-specific configuration
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (8 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 09/23] mini-os: remove per-fd evtchn limit Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 11/23] mini-os: Move test functions into test.c Daniel De Graaf
` (12 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Instead of using CONFIG_QEMU and CONFIG_GRUB to enable or disable minios
code, create CONFIG_ items for features and use application-specific
configuration files to enable or disable the features.
The configuration flags are currently added to the compiler command
line; as the number of flags grows this may need to move to a header.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
| 52 ++++++++++++++++++++++++++++++++++++++++-----
extras/mini-os/main.c | 16 +++++++-------
| 4 +-
stubdom/Makefile | 8 +++---
stubdom/grub/minios.cfg | 2 +
stubdom/ioemu-minios.cfg | 1 +
6 files changed, 63 insertions(+), 20 deletions(-)
create mode 100644 stubdom/c/minios.cfg
create mode 100644 stubdom/caml/minios.cfg
create mode 100644 stubdom/grub/minios.cfg
create mode 100644 stubdom/ioemu-minios.cfg
--git a/extras/mini-os/Makefile b/extras/mini-os/Makefile
index c2ee062..be38170 100644
--- a/extras/mini-os/Makefile
+++ b/extras/mini-os/Makefile
@@ -8,10 +8,25 @@ export XEN_ROOT = $(CURDIR)/../..
include $(XEN_ROOT)/Config.mk
OBJ_DIR ?= $(CURDIR)
-ifneq ($(stubdom),y)
+ifeq ($(MINIOS_CONFIG),)
include Config.mk
+else
+EXTRA_DEPS += $(MINIOS_CONFIG)
+include $(MINIOS_CONFIG)
endif
+# Configuration defaults
+CONFIG_START_NETWORK ?= y
+CONFIG_SPARSE_BSS ?= y
+CONFIG_QEMU_XS_ARGS ?= n
+
+# Export config items as compiler directives
+flags-$(CONFIG_START_NETWORK) += -DCONFIG_START_NETWORK
+flags-$(CONFIG_SPARSE_BSS) += -DCONFIG_SPARSE_BSS
+flags-$(CONFIG_QEMU_XS_ARGS) += -DCONFIG_QEMU_XS_ARGS
+
+DEF_CFLAGS += $(flags-y)
+
# Include common mini-os makerules.
include minios.mk
@@ -34,13 +49,38 @@ TARGET := mini-os
# Subdirectories common to mini-os
SUBDIRS := lib xenbus console
+src-y += blkfront.c
+src-y += daytime.c
+src-y += events.c
+src-y += fbfront.c
+src-y += gntmap.c
+src-y += gnttab.c
+src-y += hypervisor.c
+src-y += kernel.c
+src-y += lock.c
+src-y += main.c
+src-y += mm.c
+src-y += netfront.c
+src-y += pcifront.c
+src-y += sched.c
+
+src-y += lib/ctype.c
+src-y += lib/math.c
+src-y += lib/printf.c
+src-y += lib/stack_chk_fail.c
+src-y += lib/string.c
+src-y += lib/sys.c
+src-y += lib/xmalloc.c
+src-y += lib/xs.c
+
+src-y += xenbus/xenbus.c
+
+src-y += console/console.c
+src-y += console/xencons_ring.c
+
# The common mini-os objects to build.
APP_OBJS :=
-OBJS := $(patsubst %.c,$(OBJ_DIR)/%.o,$(wildcard *.c))
-OBJS += $(patsubst %.c,$(OBJ_DIR)/%.o,$(wildcard lib/*.c))
-OBJS += $(patsubst %.c,$(OBJ_DIR)/%.o,$(wildcard xenbus/*.c))
-OBJS += $(patsubst %.c,$(OBJ_DIR)/%.o,$(wildcard console/*.c))
-
+OBJS := $(patsubst %.c,$(OBJ_DIR)/%.o,$(src-y))
.PHONY: default
default: $(OBJ_DIR)/$(TARGET)
diff --git a/extras/mini-os/main.c b/extras/mini-os/main.c
index b95b889..aeda548 100644
--- a/extras/mini-os/main.c
+++ b/extras/mini-os/main.c
@@ -43,13 +43,13 @@ extern char __app_bss_start, __app_bss_end;
static void call_main(void *p)
{
char *c, quote;
-#ifdef CONFIG_QEMU
+#ifdef CONFIG_QEMU_XS_ARGS
char *domargs, *msg;
#endif
int argc;
char **argv;
char *envp[] = { NULL };
-#ifdef CONFIG_QEMU
+#ifdef CONFIG_QEMU_XS_ARGS
char *vm;
char path[128];
int domid;
@@ -60,15 +60,15 @@ static void call_main(void *p)
* crashing. */
//sleep(1);
-#ifndef CONFIG_GRUB
+#ifdef CONFIG_SPARSE_BSS
sparse((unsigned long) &__app_bss_start, &__app_bss_end - &__app_bss_start);
-#if defined(HAVE_LWIP) && !defined(CONFIG_QEMU)
- start_networking();
#endif
+#if defined(HAVE_LWIP) && defined(CONFIG_START_NETWORK)
+ start_networking();
#endif
create_thread("pcifront", pcifront_watches, NULL);
-#ifdef CONFIG_QEMU
+#ifdef CONFIG_QEMU_XS_ARGS
/* Fetch argc, argv from XenStore */
domid = xenbus_read_integer("target");
if (domid == -1) {
@@ -132,7 +132,7 @@ static void call_main(void *p)
#define PARSE_ARGS_STORE(ARGS) PARSE_ARGS(ARGS, argv[argc++] = c, memmove(c, c + 1, strlen(c + 1) + 1), *c++ = 0)
PARSE_ARGS_COUNT((char*)start_info.cmd_line);
-#ifdef CONFIG_QEMU
+#ifdef CONFIG_QEMU_XS_ARGS
PARSE_ARGS_COUNT(domargs);
#endif
@@ -141,7 +141,7 @@ static void call_main(void *p)
argc = 1;
PARSE_ARGS_STORE((char*)start_info.cmd_line)
-#ifdef CONFIG_QEMU
+#ifdef CONFIG_QEMU_XS_ARGS
PARSE_ARGS_STORE(domargs)
#endif
--git a/extras/mini-os/minios.mk b/extras/mini-os/minios.mk
index 698648a..48ed768 100644
--- a/extras/mini-os/minios.mk
+++ b/extras/mini-os/minios.mk
@@ -39,8 +39,8 @@ LDFLAGS := $(DEF_LDFLAGS) $(ARCH_LDFLAGS)
# Special build dependencies.
# Rebuild all after touching this/these file(s)
-EXTRA_DEPS = $(MINI-OS_ROOT)/minios.mk \
- $(MINI-OS_ROOT)/$(TARGET_ARCH_DIR)/arch.mk
+EXTRA_DEPS += $(MINI-OS_ROOT)/minios.mk
+EXTRA_DEPS += $(MINI-OS_ROOT)/$(TARGET_ARCH_DIR)/arch.mk
# Find all header files for checking dependencies.
HDRS := $(wildcard $(MINI-OS_ROOT)/include/*.h)
diff --git a/stubdom/Makefile b/stubdom/Makefile
index e9dbf02..d4da2bb 100644
--- a/stubdom/Makefile
+++ b/stubdom/Makefile
@@ -341,19 +341,19 @@ grub: grub-upstream $(CROSS_ROOT)
.PHONY: ioemu-stubdom
ioemu-stubdom: APP_OBJS=$(CURDIR)/ioemu/i386-stubdom/qemu.a $(CURDIR)/ioemu/i386-stubdom/libqemu.a $(CURDIR)/ioemu/libqemu_common.a
ioemu-stubdom: mini-os-$(XEN_TARGET_ARCH)-ioemu lwip-$(XEN_TARGET_ARCH) libxc ioemu
- DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="-DCONFIG_QEMU $(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS="$(APP_OBJS)"
+ DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/ioemu-minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS="$(APP_OBJS)"
.PHONY: caml-stubdom
caml-stubdom: mini-os-$(XEN_TARGET_ARCH)-caml lwip-$(XEN_TARGET_ARCH) libxc cross-ocaml caml
- DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="-DCONFIG_CAML $(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS="$(CURDIR)/caml/main-caml.o $(CURDIR)/caml/caml.o $(CAMLLIB)/libasmrun.a"
+ DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/caml/minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS="$(CURDIR)/caml/main-caml.o $(CURDIR)/caml/caml.o $(CAMLLIB)/libasmrun.a"
.PHONY: c-stubdom
c-stubdom: mini-os-$(XEN_TARGET_ARCH)-c lwip-$(XEN_TARGET_ARCH) libxc c
- DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="-DCONFIG_C $(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS=$(CURDIR)/c/main.a
+ DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/c/minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS=$(CURDIR)/c/main.a
.PHONY: pv-grub
pv-grub: mini-os-$(XEN_TARGET_ARCH)-grub libxc grub
- DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="-DCONFIG_GRUB $(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< APP_OBJS=$(CURDIR)/grub-$(XEN_TARGET_ARCH)/main.a
+ DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/grub/minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< APP_OBJS=$(CURDIR)/grub-$(XEN_TARGET_ARCH)/main.a
#########
# install
diff --git a/stubdom/c/minios.cfg b/stubdom/c/minios.cfg
new file mode 100644
index 0000000..e69de29
diff --git a/stubdom/caml/minios.cfg b/stubdom/caml/minios.cfg
new file mode 100644
index 0000000..e69de29
diff --git a/stubdom/grub/minios.cfg b/stubdom/grub/minios.cfg
new file mode 100644
index 0000000..40cfa68
--- /dev/null
+++ b/stubdom/grub/minios.cfg
@@ -0,0 +1,2 @@
+CONFIG_START_NETWORK=n
+CONFIG_SPARSE_BSS=n
diff --git a/stubdom/ioemu-minios.cfg b/stubdom/ioemu-minios.cfg
new file mode 100644
index 0000000..7ea1d2f
--- /dev/null
+++ b/stubdom/ioemu-minios.cfg
@@ -0,0 +1 @@
+CONFIG_QEMU_XS_ARGS=y
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 11/23] mini-os: Move test functions into test.c
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (9 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 10/23] mini-os: create app-specific configuration Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 12/23] mini-os: make frontends and xenbus optional Daniel De Graaf
` (11 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
While useful, these test functions should not be compiled into every
mini-os instance that we compile.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
| 1 +
| 414 -----------------------------------
| 139 +-----------
stubdom/c/minios.cfg | 1 +
stubdom/caml/minios.cfg | 1 +
5 files changed, 8 insertions(+), 548 deletions(-)
copy extras/mini-os/{kernel.c => test.c} (80%)
--git a/extras/mini-os/Makefile b/extras/mini-os/Makefile
index be38170..96c661a 100644
--- a/extras/mini-os/Makefile
+++ b/extras/mini-os/Makefile
@@ -63,6 +63,7 @@ src-y += mm.c
src-y += netfront.c
src-y += pcifront.c
src-y += sched.c
+src-$(CONFIG_TEST) += test.c
src-y += lib/ctype.c
src-y += lib/math.c
--git a/extras/mini-os/kernel.c b/extras/mini-os/kernel.c
index 2875bf1..c8b9b8c 100644
--- a/extras/mini-os/kernel.c
+++ b/extras/mini-os/kernel.c
@@ -46,8 +46,6 @@
#include <xen/features.h>
#include <xen/version.h>
-static struct netfront_dev *net_dev;
-
uint8_t xen_features[XENFEAT_NR_SUBMAPS * 32];
void setup_xen_features(void)
@@ -66,407 +64,10 @@ void setup_xen_features(void)
}
}
-void test_xenbus(void);
-
-static void xenbus_tester(void *p)
-{
- printk("Xenbus tests disabled, because of a Xend bug.\n");
- /* test_xenbus(); */
-}
-
-static void periodic_thread(void *p)
-{
- struct timeval tv;
- printk("Periodic thread started.\n");
- for(;;)
- {
- gettimeofday(&tv, NULL);
- printk("T(s=%ld us=%ld)\n", tv.tv_sec, tv.tv_usec);
- msleep(1000);
- }
-}
-
-static void netfront_thread(void *p)
-{
- net_dev = init_netfront(NULL, NULL, NULL, NULL);
-}
-
-static struct blkfront_dev *blk_dev;
-static struct blkfront_info blk_info;
-static uint64_t blk_size_read;
-static uint64_t blk_size_write;
-
-struct blk_req {
- struct blkfront_aiocb aiocb;
- int rand_value;
- struct blk_req *next;
-};
-
-#ifdef BLKTEST_WRITE
-static struct blk_req *blk_to_read;
-#endif
-
-static struct blk_req *blk_alloc_req(uint64_t sector)
-{
- struct blk_req *req = xmalloc(struct blk_req);
- req->aiocb.aio_dev = blk_dev;
- req->aiocb.aio_buf = _xmalloc(blk_info.sector_size, blk_info.sector_size);
- req->aiocb.aio_nbytes = blk_info.sector_size;
- req->aiocb.aio_offset = sector * blk_info.sector_size;
- req->aiocb.data = req;
- req->next = NULL;
- return req;
-}
-
-static void blk_read_completed(struct blkfront_aiocb *aiocb, int ret)
-{
- struct blk_req *req = aiocb->data;
- if (ret)
- printk("got error code %d when reading at offset %ld\n", ret, aiocb->aio_offset);
- else
- blk_size_read += blk_info.sector_size;
- free(aiocb->aio_buf);
- free(req);
-}
-
-static void blk_read_sector(uint64_t sector)
-{
- struct blk_req *req;
-
- req = blk_alloc_req(sector);
- req->aiocb.aio_cb = blk_read_completed;
-
- blkfront_aio_read(&req->aiocb);
-}
-
-#ifdef BLKTEST_WRITE
-static void blk_write_read_completed(struct blkfront_aiocb *aiocb, int ret)
-{
- struct blk_req *req = aiocb->data;
- int rand_value;
- int i;
- int *buf;
-
- if (ret) {
- printk("got error code %d when reading back at offset %ld\n", ret, aiocb->aio_offset);
- free(aiocb->aio_buf);
- free(req);
- return;
- }
- blk_size_read += blk_info.sector_size;
- buf = (int*) aiocb->aio_buf;
- rand_value = req->rand_value;
- for (i = 0; i < blk_info.sector_size / sizeof(int); i++) {
- if (buf[i] != rand_value) {
- printk("bogus data at offset %ld\n", aiocb->aio_offset + i);
- break;
- }
- rand_value *= RAND_MIX;
- }
- free(aiocb->aio_buf);
- free(req);
-}
-
-static void blk_write_completed(struct blkfront_aiocb *aiocb, int ret)
-{
- struct blk_req *req = aiocb->data;
- if (ret) {
- printk("got error code %d when writing at offset %ld\n", ret, aiocb->aio_offset);
- free(aiocb->aio_buf);
- free(req);
- return;
- }
- blk_size_write += blk_info.sector_size;
- /* Push write check */
- req->next = blk_to_read;
- blk_to_read = req;
-}
-
-static void blk_write_sector(uint64_t sector)
-{
- struct blk_req *req;
- int rand_value;
- int i;
- int *buf;
-
- req = blk_alloc_req(sector);
- req->aiocb.aio_cb = blk_write_completed;
- req->rand_value = rand_value = rand();
-
- buf = (int*) req->aiocb.aio_buf;
- for (i = 0; i < blk_info.sector_size / sizeof(int); i++) {
- buf[i] = rand_value;
- rand_value *= RAND_MIX;
- }
-
- blkfront_aio_write(&req->aiocb);
-}
-#endif
-
-static void blkfront_thread(void *p)
-{
- time_t lasttime = 0;
-
- blk_dev = init_blkfront(NULL, &blk_info);
- if (!blk_dev)
- return;
-
- if (blk_info.info & VDISK_CDROM)
- printk("Block device is a CDROM\n");
- if (blk_info.info & VDISK_REMOVABLE)
- printk("Block device is removable\n");
- if (blk_info.info & VDISK_READONLY)
- printk("Block device is read-only\n");
-
-#ifdef BLKTEST_WRITE
- if (blk_info.mode == O_RDWR) {
- blk_write_sector(0);
- blk_write_sector(blk_info.sectors-1);
- } else
-#endif
- {
- blk_read_sector(0);
- blk_read_sector(blk_info.sectors-1);
- }
-
- while (1) {
- uint64_t sector = rand() % blk_info.sectors;
- struct timeval tv;
-#ifdef BLKTEST_WRITE
- if (blk_info.mode == O_RDWR)
- blk_write_sector(sector);
- else
-#endif
- blk_read_sector(sector);
- blkfront_aio_poll(blk_dev);
- gettimeofday(&tv, NULL);
- if (tv.tv_sec > lasttime + 10) {
- printk("%llu read, %llu write\n", blk_size_read, blk_size_write);
- lasttime = tv.tv_sec;
- }
-
-#ifdef BLKTEST_WRITE
- while (blk_to_read) {
- struct blk_req *req = blk_to_read;
- blk_to_read = blk_to_read->next;
- req->aiocb.aio_cb = blk_write_read_completed;
- blkfront_aio_read(&req->aiocb);
- }
-#endif
- }
-}
-
-#define WIDTH 800
-#define HEIGHT 600
-#define DEPTH 32
-
-static uint32_t *fb;
-static int refresh_period = 50;
-static struct fbfront_dev *fb_dev;
-static struct semaphore fbfront_sem = __SEMAPHORE_INITIALIZER(fbfront_sem, 0);
-
-static void fbfront_drawvert(int x, int y1, int y2, uint32_t color)
-{
- int y;
- if (x < 0)
- return;
- if (x >= WIDTH)
- return;
- if (y1 < 0)
- y1 = 0;
- if (y2 >= HEIGHT)
- y2 = HEIGHT-1;
- for (y = y1; y <= y2; y++)
- fb[x + y*WIDTH] ^= color;
-}
-
-static void fbfront_drawhoriz(int x1, int x2, int y, uint32_t color)
-{
- int x;
- if (y < 0)
- return;
- if (y >= HEIGHT)
- return;
- if (x1 < 0)
- x1 = 0;
- if (x2 >= WIDTH)
- x2 = WIDTH-1;
- for (x = x1; x <= x2; x++)
- fb[x + y*WIDTH] ^= color;
-}
-
-static void fbfront_thread(void *p)
-{
- size_t line_length = WIDTH * (DEPTH / 8);
- size_t memsize = HEIGHT * line_length;
- unsigned long *mfns;
- int i, n = (memsize + PAGE_SIZE-1) / PAGE_SIZE;
-
- memsize = n * PAGE_SIZE;
- fb = _xmalloc(memsize, PAGE_SIZE);
- memset(fb, 0, memsize);
- mfns = xmalloc_array(unsigned long, n);
- for (i = 0; i < n; i++)
- mfns[i] = virtual_to_mfn((char *) fb + i * PAGE_SIZE);
- fb_dev = init_fbfront(NULL, mfns, WIDTH, HEIGHT, DEPTH, line_length, n);
- xfree(mfns);
- if (!fb_dev) {
- xfree(fb);
- return;
- }
- up(&fbfront_sem);
-}
-
-static void clip_cursor(int *x, int *y)
-{
- if (*x < 0)
- *x = 0;
- if (*x >= WIDTH)
- *x = WIDTH - 1;
- if (*y < 0)
- *y = 0;
- if (*y >= HEIGHT)
- *y = HEIGHT - 1;
-}
-
-static void refresh_cursor(int new_x, int new_y)
-{
- static int old_x = -1, old_y = -1;
-
- if (!refresh_period)
- return;
-
- if (old_x != -1 && old_y != -1) {
- fbfront_drawvert(old_x, old_y + 1, old_y + 8, 0xffffffff);
- fbfront_drawhoriz(old_x + 1, old_x + 8, old_y, 0xffffffff);
- fbfront_update(fb_dev, old_x, old_y, 9, 9);
- }
- old_x = new_x;
- old_y = new_y;
- fbfront_drawvert(new_x, new_y + 1, new_y + 8, 0xffffffff);
- fbfront_drawhoriz(new_x + 1, new_x + 8, new_y, 0xffffffff);
- fbfront_update(fb_dev, new_x, new_y, 9, 9);
-}
-
-static struct kbdfront_dev *kbd_dev;
-static void kbdfront_thread(void *p)
-{
- DEFINE_WAIT(w);
- DEFINE_WAIT(w2);
- int x = WIDTH / 2, y = HEIGHT / 2, z = 0;
-
- kbd_dev = init_kbdfront(NULL, 1);
- if (!kbd_dev)
- return;
-
- down(&fbfront_sem);
- refresh_cursor(x, y);
- while (1) {
- union xenkbd_in_event kbdevent;
- union xenfb_in_event fbevent;
- int sleep = 1;
-
- add_waiter(w, kbdfront_queue);
- add_waiter(w2, fbfront_queue);
-
- while (kbdfront_receive(kbd_dev, &kbdevent, 1) != 0) {
- sleep = 0;
- switch(kbdevent.type) {
- case XENKBD_TYPE_MOTION:
- printk("motion x:%d y:%d z:%d\n",
- kbdevent.motion.rel_x,
- kbdevent.motion.rel_y,
- kbdevent.motion.rel_z);
- x += kbdevent.motion.rel_x;
- y += kbdevent.motion.rel_y;
- z += kbdevent.motion.rel_z;
- clip_cursor(&x, &y);
- refresh_cursor(x, y);
- break;
- case XENKBD_TYPE_POS:
- printk("pos x:%d y:%d dz:%d\n",
- kbdevent.pos.abs_x,
- kbdevent.pos.abs_y,
- kbdevent.pos.rel_z);
- x = kbdevent.pos.abs_x;
- y = kbdevent.pos.abs_y;
- z = kbdevent.pos.rel_z;
- clip_cursor(&x, &y);
- refresh_cursor(x, y);
- break;
- case XENKBD_TYPE_KEY:
- printk("key %d %s\n",
- kbdevent.key.keycode,
- kbdevent.key.pressed ? "pressed" : "released");
- if (kbdevent.key.keycode == BTN_LEFT) {
- printk("mouse %s at (%d,%d,%d)\n",
- kbdevent.key.pressed ? "clic" : "release", x, y, z);
- if (kbdevent.key.pressed) {
- uint32_t color = rand();
- fbfront_drawvert(x - 16, y - 16, y + 15, color);
- fbfront_drawhoriz(x - 16, x + 15, y + 16, color);
- fbfront_drawvert(x + 16, y - 15, y + 16, color);
- fbfront_drawhoriz(x - 15, x + 16, y - 16, color);
- fbfront_update(fb_dev, x - 16, y - 16, 33, 33);
- }
- } else if (kbdevent.key.keycode == KEY_Q) {
- struct sched_shutdown sched_shutdown = { .reason = SHUTDOWN_poweroff };
- HYPERVISOR_sched_op(SCHEDOP_shutdown, &sched_shutdown);
- do_exit();
- }
- break;
- }
- }
- while (fbfront_receive(fb_dev, &fbevent, 1) != 0) {
- sleep = 0;
- switch(fbevent.type) {
- case XENFB_TYPE_REFRESH_PERIOD:
- refresh_period = fbevent.refresh_period.period;
- printk("refresh period %d\n", refresh_period);
- refresh_cursor(x, y);
- break;
- }
- }
- if (sleep)
- schedule();
- }
-}
-
-static struct pcifront_dev *pci_dev;
-
-static void print_pcidev(unsigned int domain, unsigned int bus, unsigned int slot, unsigned int fun)
-{
- unsigned int vendor, device, rev, class;
-
- pcifront_conf_read(pci_dev, domain, bus, slot, fun, 0x00, 2, &vendor);
- pcifront_conf_read(pci_dev, domain, bus, slot, fun, 0x02, 2, &device);
- pcifront_conf_read(pci_dev, domain, bus, slot, fun, 0x08, 1, &rev);
- pcifront_conf_read(pci_dev, domain, bus, slot, fun, 0x0a, 2, &class);
-
- printk("%04x:%02x:%02x.%02x %04x: %04x:%04x (rev %02x)\n", domain, bus, slot, fun, class, vendor, device, rev);
-}
-
-static void pcifront_thread(void *p)
-{
- pcifront_watches(NULL);
- pci_dev = init_pcifront(NULL);
- if (!pci_dev)
- return;
- printk("PCI devices:\n");
- pcifront_scan(pci_dev, print_pcidev);
-}
-
/* This should be overridden by the application we are linked against. */
__attribute__((weak)) int app_main(start_info_t *si)
{
printk("Dummy main: start_info=%p\n", si);
- create_thread("xenbus_tester", xenbus_tester, si);
- create_thread("periodic_thread", periodic_thread, si);
- create_thread("netfront", netfront_thread, si);
- create_thread("blkfront", blkfront_thread, si);
- create_thread("fbfront", fbfront_thread, si);
- create_thread("kbdfront", kbdfront_thread, si);
- create_thread("pcifront", pcifront_thread, si);
return 0;
}
@@ -534,21 +135,6 @@ void start_kernel(start_info_t *si)
void stop_kernel(void)
{
- if (net_dev)
- shutdown_netfront(net_dev);
-
- if (blk_dev)
- shutdown_blkfront(blk_dev);
-
- if (fb_dev)
- shutdown_fbfront(fb_dev);
-
- if (kbd_dev)
- shutdown_kbdfront(kbd_dev);
-
- if (pci_dev)
- shutdown_pcifront(pci_dev);
-
/* TODO: fs import */
local_irq_disable();
--git a/extras/mini-os/kernel.c b/extras/mini-os/test.c
similarity index 80%
copy from extras/mini-os/kernel.c
copy to extras/mini-os/test.c
index 2875bf1..9039cb3 100644
--- a/extras/mini-os/kernel.c
+++ b/extras/mini-os/test.c
@@ -1,8 +1,7 @@
/******************************************************************************
- * kernel.c
+ * test.c
*
- * Assorted crap goes here, including the initial C entry point, jumped at
- * from head.S.
+ * Test code for all the various frontends; split from kernel.c
*
* Copyright (c) 2002-2003, K A Fraser & R Neugebauer
* Copyright (c) 2005, Grzegorz Milos, Intel Research Cambridge
@@ -48,24 +47,6 @@
static struct netfront_dev *net_dev;
-uint8_t xen_features[XENFEAT_NR_SUBMAPS * 32];
-
-void setup_xen_features(void)
-{
- xen_feature_info_t fi;
- int i, j;
-
- for (i = 0; i < XENFEAT_NR_SUBMAPS; i++)
- {
- fi.submap_idx = i;
- if (HYPERVISOR_xen_version(XENVER_get_features, &fi) < 0)
- break;
-
- for (j=0; j<32; j++)
- xen_features[i*32+j] = !!(fi.submap & 1<<j);
- }
-}
-
void test_xenbus(void);
static void xenbus_tester(void *p)
@@ -456,10 +437,9 @@ static void pcifront_thread(void *p)
pcifront_scan(pci_dev, print_pcidev);
}
-/* This should be overridden by the application we are linked against. */
-__attribute__((weak)) int app_main(start_info_t *si)
+int app_main(start_info_t *si)
{
- printk("Dummy main: start_info=%p\n", si);
+ printk("Test main: start_info=%p\n", si);
create_thread("xenbus_tester", xenbus_tester, si);
create_thread("periodic_thread", periodic_thread, si);
create_thread("netfront", netfront_thread, si);
@@ -470,69 +450,7 @@ __attribute__((weak)) int app_main(start_info_t *si)
return 0;
}
-/*
- * INITIAL C ENTRY POINT.
- */
-void start_kernel(start_info_t *si)
-{
- static char hello[] = "Bootstrapping...\n";
-
- (void)HYPERVISOR_console_io(CONSOLEIO_write, strlen(hello), hello);
-
- arch_init(si);
-
- trap_init();
-
- /* print out some useful information */
- printk("Xen Minimal OS!\n");
- printk(" start_info: %p(VA)\n", si);
- printk(" nr_pages: 0x%lx\n", si->nr_pages);
- printk(" shared_inf: 0x%08lx(MA)\n", si->shared_info);
- printk(" pt_base: %p(VA)\n", (void *)si->pt_base);
- printk("nr_pt_frames: 0x%lx\n", si->nr_pt_frames);
- printk(" mfn_list: %p(VA)\n", (void *)si->mfn_list);
- printk(" mod_start: 0x%lx(VA)\n", si->mod_start);
- printk(" mod_len: %lu\n", si->mod_len);
- printk(" flags: 0x%x\n", (unsigned int)si->flags);
- printk(" cmd_line: %s\n",
- si->cmd_line ? (const char *)si->cmd_line : "NULL");
-
- /* Set up events. */
- init_events();
-
- /* ENABLE EVENT DELIVERY. This is disabled at start of day. */
- __sti();
-
- arch_print_info();
-
- setup_xen_features();
-
- /* Init memory management. */
- init_mm();
-
- /* Init time and timers. */
- init_time();
-
- /* Init the console driver. */
- init_console();
-
- /* Init grant tables */
- init_gnttab();
-
- /* Init scheduler. */
- init_sched();
-
- /* Init XenBus */
- init_xenbus();
-
- /* Call (possibly overridden) app_main() */
- app_main(&start_info);
-
- /* Everything initialised, start idle thread */
- run_idle_thread();
-}
-
-void stop_kernel(void)
+void shutdown_frontends(void)
{
if (net_dev)
shutdown_netfront(net_dev);
@@ -548,51 +466,4 @@ void stop_kernel(void)
if (pci_dev)
shutdown_pcifront(pci_dev);
-
- /* TODO: fs import */
-
- local_irq_disable();
-
- /* Reset grant tables */
- fini_gnttab();
-
- /* Reset the console driver. */
- fini_console(NULL);
- /* TODO: record new ring mfn & event in start_info */
-
- /* Reset XenBus */
- fini_xenbus();
-
- /* Reset timers */
- fini_time();
-
- /* Reset memory management. */
- fini_mm();
-
- /* Reset events. */
- fini_events();
-
- /* Reset traps */
- trap_fini();
-
- /* Reset arch details */
- arch_fini();
-}
-
-/*
- * do_exit: This is called whenever an IRET fails in entry.S.
- * This will generally be because an application has got itself into
- * a really bad state (probably a bad CS or SS). It must be killed.
- * Of course, minimal OS doesn't have applications :-)
- */
-
-void do_exit(void)
-{
- printk("Do_exit called!\n");
- stack_walk();
- for( ;; )
- {
- struct sched_shutdown sched_shutdown = { .reason = SHUTDOWN_crash };
- HYPERVISOR_sched_op(SCHEDOP_shutdown, &sched_shutdown);
- }
}
diff --git a/stubdom/c/minios.cfg b/stubdom/c/minios.cfg
index e69de29..e1faee5 100644
--- a/stubdom/c/minios.cfg
+++ b/stubdom/c/minios.cfg
@@ -0,0 +1 @@
+CONFIG_TEST=y
diff --git a/stubdom/caml/minios.cfg b/stubdom/caml/minios.cfg
index e69de29..e1faee5 100644
--- a/stubdom/caml/minios.cfg
+++ b/stubdom/caml/minios.cfg
@@ -0,0 +1 @@
+CONFIG_TEST=y
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 12/23] mini-os: make frontends and xenbus optional
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (10 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 11/23] mini-os: Move test functions into test.c Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 13/23] mini-os: fix list.h include guard name Daniel De Graaf
` (10 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
This adds compile-time logic to disable certain frontends in mini-os:
- pcifront is disabled by default, enabled for ioemu
- blkfront, netfront, fbfront, kbdfront, consfront are enabled by default
- xenbus is required for any frontend, and is enabled by default
If all frontends and xenbus are disabled, mini-os will run without
needing to communicate with xenstore, making it suitable to run the
xenstore daemon. The console frontend is not required for the initial
console, only consoles opened via openpt or ptmx.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
| 32 +++-
| 5 -
| 2 +
.../mini-os/console/{xencons_ring.c => xenbus.c} | 182 +-------------------
| 182 +-------------------
| 2 +
| 12 ++
| 4 -
| 43 +++++
extras/mini-os/main.c | 6 +-
stubdom/ioemu-minios.cfg | 1 +
11 files changed, 96 insertions(+), 375 deletions(-)
create mode 100644 extras/mini-os/console/console.h
copy extras/mini-os/console/{xencons_ring.c => xenbus.c} (56%)
--git a/extras/mini-os/Makefile b/extras/mini-os/Makefile
index 96c661a..ea5d0b5 100644
--- a/extras/mini-os/Makefile
+++ b/extras/mini-os/Makefile
@@ -19,11 +19,25 @@ endif
CONFIG_START_NETWORK ?= y
CONFIG_SPARSE_BSS ?= y
CONFIG_QEMU_XS_ARGS ?= n
+CONFIG_PCIFRONT ?= n
+CONFIG_BLKFRONT ?= y
+CONFIG_NETFRONT ?= y
+CONFIG_FBFRONT ?= y
+CONFIG_KBDFRONT ?= y
+CONFIG_CONSFRONT ?= y
+CONFIG_XENBUS ?= y
# Export config items as compiler directives
flags-$(CONFIG_START_NETWORK) += -DCONFIG_START_NETWORK
flags-$(CONFIG_SPARSE_BSS) += -DCONFIG_SPARSE_BSS
flags-$(CONFIG_QEMU_XS_ARGS) += -DCONFIG_QEMU_XS_ARGS
+flags-$(CONFIG_PCIFRONT) += -DCONFIG_PCIFRONT
+flags-$(CONFIG_BLKFRONT) += -DCONFIG_BLKFRONT
+flags-$(CONFIG_NETFRONT) += -DCONFIG_NETFRONT
+flags-$(CONFIG_KBDFRONT) += -DCONFIG_KBDFRONT
+flags-$(CONFIG_FBFRONT) += -DCONFIG_FBFRONT
+flags-$(CONFIG_CONSFRONT) += -DCONFIG_CONSFRONT
+flags-$(CONFIG_XENBUS) += -DCONFIG_XENBUS
DEF_CFLAGS += $(flags-y)
@@ -49,10 +63,10 @@ TARGET := mini-os
# Subdirectories common to mini-os
SUBDIRS := lib xenbus console
-src-y += blkfront.c
+src-$(CONFIG_BLKFRONT) += blkfront.c
src-y += daytime.c
src-y += events.c
-src-y += fbfront.c
+src-$(CONFIG_FBFRONT) += fbfront.c
src-y += gntmap.c
src-y += gnttab.c
src-y += hypervisor.c
@@ -60,8 +74,8 @@ src-y += kernel.c
src-y += lock.c
src-y += main.c
src-y += mm.c
-src-y += netfront.c
-src-y += pcifront.c
+src-$(CONFIG_NETFRONT) += netfront.c
+src-$(CONFIG_PCIFRONT) += pcifront.c
src-y += sched.c
src-$(CONFIG_TEST) += test.c
@@ -72,12 +86,13 @@ src-y += lib/stack_chk_fail.c
src-y += lib/string.c
src-y += lib/sys.c
src-y += lib/xmalloc.c
-src-y += lib/xs.c
+src-$(CONFIG_XENBUS) += lib/xs.c
-src-y += xenbus/xenbus.c
+src-$(CONFIG_XENBUS) += xenbus/xenbus.c
src-y += console/console.c
src-y += console/xencons_ring.c
+src-$(CONFIG_CONSFRONT) += console/xenbus.c
# The common mini-os objects to build.
APP_OBJS :=
@@ -108,7 +123,10 @@ ifeq ($(lwip),y)
LWC := $(shell find $(LWIPDIR)/ -type f -name '*.c')
LWC := $(filter-out %6.c %ip6_addr.c %ethernetif.c, $(LWC))
LWO := $(patsubst %.c,%.o,$(LWC))
-LWO += $(addprefix $(OBJ_DIR)/,lwip-arch.o lwip-net.o)
+LWO += $(OBJ_DIR)/lwip-arch.o
+ifeq ($(CONFIG_NETFRONT),y)
+LWO += $(OBJ_DIR)/lwip-net.o
+endif
$(OBJ_DIR)/lwip.a: $(LWO)
$(RM) $@
--git a/extras/mini-os/console/console.c b/extras/mini-os/console/console.c
index 2b4ed9f..fec3791 100644
--- a/extras/mini-os/console/console.c
+++ b/extras/mini-os/console/console.c
@@ -158,8 +158,3 @@ void init_console(void)
/* This is also required to notify the daemon */
printk("done.\n");
}
-
-void fini_console(struct consfront_dev *dev)
-{
- if (dev) free_consfront(dev);
-}
--git a/extras/mini-os/console/console.h b/extras/mini-os/console/console.h
new file mode 100644
index 0000000..e85147a
--- /dev/null
+++ b/extras/mini-os/console/console.h
@@ -0,0 +1,2 @@
+
+void console_handle_input(evtchn_port_t port, struct pt_regs *regs, void *data);
diff --git a/extras/mini-os/console/xencons_ring.c b/extras/mini-os/console/xenbus.c
similarity index 56%
copy from extras/mini-os/console/xencons_ring.c
copy to extras/mini-os/console/xenbus.c
index af0afed..a7c517d 100644
--- a/extras/mini-os/console/xencons_ring.c
+++ b/extras/mini-os/console/xenbus.c
@@ -11,181 +11,7 @@
#include <xen/io/ring.h>
#include <mini-os/xmalloc.h>
#include <mini-os/gnttab.h>
-
-DECLARE_WAIT_QUEUE_HEAD(console_queue);
-
-static inline void notify_daemon(struct consfront_dev *dev)
-{
- /* Use evtchn: this is called early, before irq is set up. */
- if (!dev)
- notify_remote_via_evtchn(start_info.console.domU.evtchn);
- else
- notify_remote_via_evtchn(dev->evtchn);
-}
-
-static inline struct xencons_interface *xencons_interface(void)
-{
- if (start_info.console.domU.evtchn)
- return mfn_to_virt(start_info.console.domU.mfn);
- else
- return NULL;
-}
-
-int xencons_ring_send_no_notify(struct consfront_dev *dev, const char *data, unsigned len)
-{
- int sent = 0;
- struct xencons_interface *intf;
- XENCONS_RING_IDX cons, prod;
-
- if (!dev)
- intf = xencons_interface();
- else
- intf = dev->ring;
- if (!intf)
- return sent;
-
- cons = intf->out_cons;
- prod = intf->out_prod;
- mb();
- BUG_ON((prod - cons) > sizeof(intf->out));
-
- while ((sent < len) && ((prod - cons) < sizeof(intf->out)))
- intf->out[MASK_XENCONS_IDX(prod++, intf->out)] = data[sent++];
-
- wmb();
- intf->out_prod = prod;
-
- return sent;
-}
-
-int xencons_ring_send(struct consfront_dev *dev, const char *data, unsigned len)
-{
- int sent;
-
- sent = xencons_ring_send_no_notify(dev, data, len);
- notify_daemon(dev);
-
- return sent;
-}
-
-
-
-static void handle_input(evtchn_port_t port, struct pt_regs *regs, void *data)
-{
- struct consfront_dev *dev = (struct consfront_dev *) data;
-#ifdef HAVE_LIBC
- int fd = dev ? dev->fd : -1;
-
- if (fd != -1)
- files[fd].read = 1;
-
- wake_up(&console_queue);
-#else
- struct xencons_interface *intf = xencons_interface();
- XENCONS_RING_IDX cons, prod;
-
- cons = intf->in_cons;
- prod = intf->in_prod;
- mb();
- BUG_ON((prod - cons) > sizeof(intf->in));
-
- while (cons != prod) {
- xencons_rx(intf->in+MASK_XENCONS_IDX(cons,intf->in), 1, regs);
- cons++;
- }
-
- mb();
- intf->in_cons = cons;
-
- notify_daemon(dev);
-
- xencons_tx();
-#endif
-}
-
-#ifdef HAVE_LIBC
-int xencons_ring_avail(struct consfront_dev *dev)
-{
- struct xencons_interface *intf;
- XENCONS_RING_IDX cons, prod;
-
- if (!dev)
- intf = xencons_interface();
- else
- intf = dev->ring;
-
- cons = intf->in_cons;
- prod = intf->in_prod;
- mb();
- BUG_ON((prod - cons) > sizeof(intf->in));
-
- return prod - cons;
-}
-
-int xencons_ring_recv(struct consfront_dev *dev, char *data, unsigned len)
-{
- struct xencons_interface *intf;
- XENCONS_RING_IDX cons, prod;
- unsigned filled = 0;
-
- if (!dev)
- intf = xencons_interface();
- else
- intf = dev->ring;
-
- cons = intf->in_cons;
- prod = intf->in_prod;
- mb();
- BUG_ON((prod - cons) > sizeof(intf->in));
-
- while (filled < len && cons + filled != prod) {
- data[filled] = *(intf->in + MASK_XENCONS_IDX(cons + filled, intf->in));
- filled++;
- }
-
- mb();
- intf->in_cons = cons + filled;
-
- notify_daemon(dev);
-
- return filled;
-}
-#endif
-
-struct consfront_dev *xencons_ring_init(void)
-{
- int err;
- struct consfront_dev *dev;
-
- if (!start_info.console.domU.evtchn)
- return 0;
-
- dev = malloc(sizeof(struct consfront_dev));
- memset(dev, 0, sizeof(struct consfront_dev));
- dev->nodename = "device/console";
- dev->dom = 0;
- dev->backend = 0;
- dev->ring_ref = 0;
-
-#ifdef HAVE_LIBC
- dev->fd = -1;
-#endif
- dev->evtchn = start_info.console.domU.evtchn;
- dev->ring = (struct xencons_interface *) mfn_to_virt(start_info.console.domU.mfn);
-
- err = bind_evtchn(dev->evtchn, handle_input, dev);
- if (err <= 0) {
- printk("XEN console request chn bind failed %i\n", err);
- free(dev);
- return NULL;
- }
- unmask_evtchn(dev->evtchn);
-
- /* In case we have in-flight data after save/restore... */
- notify_daemon(dev);
-
- return dev;
-}
+#include "console.h"
void free_consfront(struct consfront_dev *dev)
{
@@ -262,7 +88,7 @@ struct consfront_dev *init_consfront(char *_nodename)
return NULL;
else
dev->dom = res;
- evtchn_alloc_unbound(dev->dom, handle_input, dev, &dev->evtchn);
+ evtchn_alloc_unbound(dev->dom, console_handle_input, dev, &dev->evtchn);
dev->ring = (struct xencons_interface *) alloc_page();
memset(dev->ring, 0, PAGE_SIZE);
@@ -364,8 +190,8 @@ error:
return NULL;
}
-void xencons_resume(void)
+void fini_console(struct consfront_dev *dev)
{
- (void)xencons_ring_init();
+ if (dev) free_consfront(dev);
}
--git a/extras/mini-os/console/xencons_ring.c b/extras/mini-os/console/xencons_ring.c
index af0afed..81c8e99 100644
--- a/extras/mini-os/console/xencons_ring.c
+++ b/extras/mini-os/console/xencons_ring.c
@@ -11,6 +11,7 @@
#include <xen/io/ring.h>
#include <mini-os/xmalloc.h>
#include <mini-os/gnttab.h>
+#include "console.h"
DECLARE_WAIT_QUEUE_HEAD(console_queue);
@@ -70,7 +71,7 @@ int xencons_ring_send(struct consfront_dev *dev, const char *data, unsigned len)
-static void handle_input(evtchn_port_t port, struct pt_regs *regs, void *data)
+void console_handle_input(evtchn_port_t port, struct pt_regs *regs, void *data)
{
struct consfront_dev *dev = (struct consfront_dev *) data;
#ifdef HAVE_LIBC
@@ -173,7 +174,7 @@ struct consfront_dev *xencons_ring_init(void)
dev->evtchn = start_info.console.domU.evtchn;
dev->ring = (struct xencons_interface *) mfn_to_virt(start_info.console.domU.mfn);
- err = bind_evtchn(dev->evtchn, handle_input, dev);
+ err = bind_evtchn(dev->evtchn, console_handle_input, dev);
if (err <= 0) {
printk("XEN console request chn bind failed %i\n", err);
free(dev);
@@ -187,183 +188,6 @@ struct consfront_dev *xencons_ring_init(void)
return dev;
}
-void free_consfront(struct consfront_dev *dev)
-{
- char* err = NULL;
- XenbusState state;
-
- char path[strlen(dev->backend) + 1 + 5 + 1];
- char nodename[strlen(dev->nodename) + 1 + 5 + 1];
-
- snprintf(path, sizeof(path), "%s/state", dev->backend);
- snprintf(nodename, sizeof(nodename), "%s/state", dev->nodename);
-
- if ((err = xenbus_switch_state(XBT_NIL, nodename, XenbusStateClosing)) != NULL) {
- printk("free_consfront: error changing state to %d: %s\n",
- XenbusStateClosing, err);
- goto close;
- }
- state = xenbus_read_integer(path);
- while (err == NULL && state < XenbusStateClosing)
- err = xenbus_wait_for_state_change(path, &state, &dev->events);
- if (err) free(err);
-
- if ((err = xenbus_switch_state(XBT_NIL, nodename, XenbusStateClosed)) != NULL) {
- printk("free_consfront: error changing state to %d: %s\n",
- XenbusStateClosed, err);
- goto close;
- }
-
-close:
- if (err) free(err);
- xenbus_unwatch_path_token(XBT_NIL, path, path);
-
- mask_evtchn(dev->evtchn);
- unbind_evtchn(dev->evtchn);
- free(dev->backend);
- free(dev->nodename);
-
- gnttab_end_access(dev->ring_ref);
-
- free_page(dev->ring);
- free(dev);
-}
-
-struct consfront_dev *init_consfront(char *_nodename)
-{
- xenbus_transaction_t xbt;
- char* err;
- char* message=NULL;
- int retry=0;
- char* msg = NULL;
- char nodename[256];
- char path[256];
- static int consfrontends = 3;
- struct consfront_dev *dev;
- int res;
-
- if (!_nodename)
- snprintf(nodename, sizeof(nodename), "device/console/%d", consfrontends);
- else
- strncpy(nodename, _nodename, sizeof(nodename));
-
- printk("******************* CONSFRONT for %s **********\n\n\n", nodename);
-
- consfrontends++;
- dev = malloc(sizeof(*dev));
- memset(dev, 0, sizeof(*dev));
- dev->nodename = strdup(nodename);
-#ifdef HAVE_LIBC
- dev->fd = -1;
-#endif
-
- snprintf(path, sizeof(path), "%s/backend-id", nodename);
- if ((res = xenbus_read_integer(path)) < 0)
- return NULL;
- else
- dev->dom = res;
- evtchn_alloc_unbound(dev->dom, handle_input, dev, &dev->evtchn);
-
- dev->ring = (struct xencons_interface *) alloc_page();
- memset(dev->ring, 0, PAGE_SIZE);
- dev->ring_ref = gnttab_grant_access(dev->dom, virt_to_mfn(dev->ring), 0);
-
- dev->events = NULL;
-
-again:
- err = xenbus_transaction_start(&xbt);
- if (err) {
- printk("starting transaction\n");
- free(err);
- }
-
- err = xenbus_printf(xbt, nodename, "ring-ref","%u",
- dev->ring_ref);
- if (err) {
- message = "writing ring-ref";
- goto abort_transaction;
- }
- err = xenbus_printf(xbt, nodename,
- "port", "%u", dev->evtchn);
- if (err) {
- message = "writing event-channel";
- goto abort_transaction;
- }
- err = xenbus_printf(xbt, nodename,
- "protocol", "%s", XEN_IO_PROTO_ABI_NATIVE);
- if (err) {
- message = "writing protocol";
- goto abort_transaction;
- }
-
- err = xenbus_printf(xbt, nodename, "type", "%s", "ioemu");
- if (err) {
- message = "writing type";
- goto abort_transaction;
- }
-
- snprintf(path, sizeof(path), "%s/state", nodename);
- err = xenbus_switch_state(xbt, path, XenbusStateConnected);
- if (err) {
- message = "switching state";
- goto abort_transaction;
- }
-
-
- err = xenbus_transaction_end(xbt, 0, &retry);
- if (err) free(err);
- if (retry) {
- goto again;
- printk("completing transaction\n");
- }
-
- goto done;
-
-abort_transaction:
- free(err);
- err = xenbus_transaction_end(xbt, 1, &retry);
- goto error;
-
-done:
-
- snprintf(path, sizeof(path), "%s/backend", nodename);
- msg = xenbus_read(XBT_NIL, path, &dev->backend);
- if (msg) {
- printk("Error %s when reading the backend path %s\n", msg, path);
- goto error;
- }
-
- printk("backend at %s\n", dev->backend);
-
- {
- XenbusState state;
- char path[strlen(dev->backend) + 1 + 19 + 1];
- snprintf(path, sizeof(path), "%s/state", dev->backend);
-
- xenbus_watch_path_token(XBT_NIL, path, path, &dev->events);
- msg = NULL;
- state = xenbus_read_integer(path);
- while (msg == NULL && state < XenbusStateConnected)
- msg = xenbus_wait_for_state_change(path, &state, &dev->events);
- if (msg != NULL || state != XenbusStateConnected) {
- printk("backend not available, state=%d\n", state);
- xenbus_unwatch_path_token(XBT_NIL, path, path);
- goto error;
- }
- }
- unmask_evtchn(dev->evtchn);
-
- printk("**************************\n");
-
- return dev;
-
-error:
- free(msg);
- free(err);
- free_consfront(dev);
- return NULL;
-}
-
void xencons_resume(void)
{
(void)xencons_ring_init();
--git a/extras/mini-os/include/lib.h b/extras/mini-os/include/lib.h
index 12070c3..9c69440 100644
--- a/extras/mini-os/include/lib.h
+++ b/extras/mini-os/include/lib.h
@@ -182,11 +182,13 @@ extern struct file {
struct {
struct consfront_dev *dev;
} cons;
+#ifdef CONFIG_XENBUS
struct {
/* To each xenbus FD is associated a queue of watch events for this
* FD. */
xenbus_event_queue events;
} xenbus;
+#endif
};
int read; /* maybe available for read */
} files[];
--git a/extras/mini-os/include/xenbus.h b/extras/mini-os/include/xenbus.h
index de618fc..cd6316e 100644
--- a/extras/mini-os/include/xenbus.h
+++ b/extras/mini-os/include/xenbus.h
@@ -6,8 +6,14 @@
typedef unsigned long xenbus_transaction_t;
#define XBT_NIL ((xenbus_transaction_t)0)
+#ifdef CONFIG_XENBUS
/* Initialize the XenBus system. */
void init_xenbus(void);
+#else
+static inline void init_xenbus(void)
+{
+}
+#endif
/* Read the value associated with a path. Returns a malloc'd error
string on failure and sets *value to NULL. On success, *value is
@@ -98,7 +104,13 @@ char* xenbus_printf(xenbus_transaction_t xbt,
/* Utility function to figure out our domain id */
domid_t xenbus_get_self_id(void);
+#ifdef CONFIG_XENBUS
/* Reset the XenBus system. */
void fini_xenbus(void);
+#else
+static inline void fini_xenbus(void)
+{
+}
+#endif
#endif /* XENBUS_H__ */
--git a/extras/mini-os/kernel.c b/extras/mini-os/kernel.c
index c8b9b8c..a8dfe6d 100644
--- a/extras/mini-os/kernel.c
+++ b/extras/mini-os/kernel.c
@@ -142,10 +142,6 @@ void stop_kernel(void)
/* Reset grant tables */
fini_gnttab();
- /* Reset the console driver. */
- fini_console(NULL);
- /* TODO: record new ring mfn & event in start_info */
-
/* Reset XenBus */
fini_xenbus();
--git a/extras/mini-os/lib/sys.c b/extras/mini-os/lib/sys.c
index b7b3aff..dcdf096 100644
--- a/extras/mini-os/lib/sys.c
+++ b/extras/mini-os/lib/sys.c
@@ -164,6 +164,7 @@ int mkdir(const char *pathname, mode_t mode)
return -1;
}
+#ifdef CONFIG_CONSFRONT
int posix_openpt(int flags)
{
struct consfront_dev *dev;
@@ -192,6 +193,18 @@ int open_savefile(const char *path, int save)
printk("fd(%d) = open_savefile\n", dev->fd);
return(dev->fd);
}
+#else
+int posix_openpt(int flags)
+{
+ errno = EIO;
+ return -1;
+}
+int open_savefile(const char *path, int save)
+{
+ errno = EIO;
+ return -1;
+}
+#endif
int open(const char *pathname, int flags, ...)
{
@@ -241,6 +254,7 @@ int read(int fd, void *buf, size_t nbytes)
case FTYPE_SOCKET:
return lwip_read(files[fd].socket.fd, buf, nbytes);
#endif
+#ifdef CONFIG_NETFRONT
case FTYPE_TAP: {
ssize_t ret;
ret = netfront_receive(files[fd].tap.dev, buf, nbytes);
@@ -250,6 +264,8 @@ int read(int fd, void *buf, size_t nbytes)
}
return ret;
}
+#endif
+#ifdef CONFIG_KBDFRONT
case FTYPE_KBD: {
int ret, n;
n = nbytes / sizeof(union xenkbd_in_event);
@@ -260,6 +276,8 @@ int read(int fd, void *buf, size_t nbytes)
}
return ret * sizeof(union xenkbd_in_event);
}
+#endif
+#ifdef CONFIG_FBFRONT
case FTYPE_FB: {
int ret, n;
n = nbytes / sizeof(union xenfb_in_event);
@@ -270,6 +288,7 @@ int read(int fd, void *buf, size_t nbytes)
}
return ret * sizeof(union xenfb_in_event);
}
+#endif
default:
break;
}
@@ -297,9 +316,11 @@ int write(int fd, const void *buf, size_t nbytes)
case FTYPE_SOCKET:
return lwip_write(files[fd].socket.fd, (void*) buf, nbytes);
#endif
+#ifdef CONFIG_NETFRONT
case FTYPE_TAP:
netfront_xmit(files[fd].tap.dev, (void*) buf, nbytes);
return nbytes;
+#endif
default:
break;
}
@@ -326,9 +347,11 @@ int close(int fd)
default:
files[fd].type = FTYPE_NONE;
return 0;
+#ifdef CONFIG_XENBUS
case FTYPE_XENBUS:
xs_daemon_close((void*)(intptr_t) fd);
return 0;
+#endif
#ifdef HAVE_LWIP
case FTYPE_SOCKET: {
int res = lwip_close(files[fd].socket.fd);
@@ -345,27 +368,37 @@ int close(int fd)
case FTYPE_GNTMAP:
minios_gnttab_close_fd(fd);
return 0;
+#ifdef CONFIG_NETFRONT
case FTYPE_TAP:
shutdown_netfront(files[fd].tap.dev);
files[fd].type = FTYPE_NONE;
return 0;
+#endif
+#ifdef CONFIG_BLKFRONT
case FTYPE_BLK:
shutdown_blkfront(files[fd].blk.dev);
files[fd].type = FTYPE_NONE;
return 0;
+#endif
+#ifdef CONFIG_KBDFRONT
case FTYPE_KBD:
shutdown_kbdfront(files[fd].kbd.dev);
files[fd].type = FTYPE_NONE;
return 0;
+#endif
+#ifdef CONFIG_FBFRONT
case FTYPE_FB:
shutdown_fbfront(files[fd].fb.dev);
files[fd].type = FTYPE_NONE;
return 0;
+#endif
+#ifdef CONFIG_CONSFRONT
case FTYPE_SAVEFILE:
case FTYPE_CONSOLE:
fini_console(files[fd].cons.dev);
files[fd].type = FTYPE_NONE;
return 0;
+#endif
case FTYPE_NONE:
break;
}
@@ -611,6 +644,7 @@ static int select_poll(int nfds, fd_set *readfds, fd_set *writefds, fd_set *exce
n++;
FD_CLR(i, exceptfds);
break;
+#ifdef CONFIG_XENBUS
case FTYPE_XENBUS:
if (FD_ISSET(i, readfds)) {
if (files[i].xenbus.events)
@@ -621,6 +655,7 @@ static int select_poll(int nfds, fd_set *readfds, fd_set *writefds, fd_set *exce
FD_CLR(i, writefds);
FD_CLR(i, exceptfds);
break;
+#endif
case FTYPE_EVTCHN:
case FTYPE_TAP:
case FTYPE_BLK:
@@ -727,11 +762,19 @@ int select(int nfds, fd_set *readfds, fd_set *writefds, fd_set *exceptfds,
/* Tell people we're going to sleep before looking at what they are
* saying, hence letting them wake us if events happen between here and
* schedule() */
+#ifdef CONFIG_NETFRONT
add_waiter(w1, netfront_queue);
+#endif
add_waiter(w2, event_queue);
+#ifdef CONFIG_BLKFRONT
add_waiter(w3, blkfront_queue);
+#endif
+#ifdef CONFIG_XENBUS
add_waiter(w4, xenbus_watch_queue);
+#endif
+#ifdef CONFIG_KBDFRONT
add_waiter(w5, kbdfront_queue);
+#endif
add_waiter(w6, console_queue);
if (readfds)
diff --git a/extras/mini-os/main.c b/extras/mini-os/main.c
index aeda548..73eb6fb 100644
--- a/extras/mini-os/main.c
+++ b/extras/mini-os/main.c
@@ -63,10 +63,12 @@ static void call_main(void *p)
#ifdef CONFIG_SPARSE_BSS
sparse((unsigned long) &__app_bss_start, &__app_bss_end - &__app_bss_start);
#endif
-#if defined(HAVE_LWIP) && defined(CONFIG_START_NETWORK)
+#if defined(HAVE_LWIP) && defined(CONFIG_START_NETWORK) && defined(CONFIG_NETFRONT)
start_networking();
#endif
+#ifdef CONFIG_PCIFRONT
create_thread("pcifront", pcifront_watches, NULL);
+#endif
#ifdef CONFIG_QEMU_XS_ARGS
/* Fetch argc, argv from XenStore */
@@ -169,7 +171,7 @@ void _exit(int ret)
close_all_files();
__libc_fini_array();
printk("main returned %d\n", ret);
-#ifdef HAVE_LWIP
+#if defined(HAVE_LWIP) && defined(CONFIG_NETFRONT)
stop_networking();
#endif
stop_kernel();
diff --git a/stubdom/ioemu-minios.cfg b/stubdom/ioemu-minios.cfg
index 7ea1d2f..e3a96da 100644
--- a/stubdom/ioemu-minios.cfg
+++ b/stubdom/ioemu-minios.cfg
@@ -1 +1,2 @@
CONFIG_QEMU_XS_ARGS=y
+CONFIG_PCIFRONT=y
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 13/23] mini-os: fix list.h include guard name
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (11 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 12/23] mini-os: make frontends and xenbus optional Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 14/23] xenstored: use grant references instead of map_foreign_range Daniel De Graaf
` (9 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
The symbol _LINUX_LIST_H collides with other header files.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
| 6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
--git a/extras/mini-os/include/list.h b/extras/mini-os/include/list.h
index a60ae23..4e6a2ac 100644
--- a/extras/mini-os/include/list.h
+++ b/extras/mini-os/include/list.h
@@ -1,5 +1,5 @@
-#ifndef _LINUX_LIST_H
-#define _LINUX_LIST_H
+#ifndef _MINIOS_LIST_H
+#define _MINIOS_LIST_H
/*
* Simple doubly linked list implementation.
@@ -186,5 +186,5 @@ static __inline__ void minios_list_splice(struct minios_list_head *list, struct
n = minios_list_entry(pos->member.next, typeof(*pos), member); \
&pos->member != (head); \
pos = n, n = minios_list_entry(n->member.next, typeof(*n), member))
-#endif /* _LINUX_LIST_H */
+#endif /* _MINIOS_LIST_H */
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 14/23] xenstored: use grant references instead of map_foreign_range
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (12 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 13/23] mini-os: fix list.h include guard name Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 15/23] xenstored: refactor socket setup code Daniel De Graaf
` (8 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
make xenstored use grantref rather than map_foreign_range (which can
only be used by privileged domains)
This patch modifies the xenstore daemon to use xc_gnttab_map_grant_ref
instead of xc_map_foreign_range where available.
Previous versions of this patch have been sent to xen-devel. See
http://lists.xensource.com/archives/html/xen-devel/2008-07/msg00610.html
http://lists.xensource.com/archives/html/xen-devel/2009-03/msg01492.html
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/xenstore/xenstored_domain.c | 52 ++++++++++++++++++++++++++++++++----
1 files changed, 46 insertions(+), 6 deletions(-)
diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
index 443af82..661d955 100644
--- a/tools/xenstore/xenstored_domain.c
+++ b/tools/xenstore/xenstored_domain.c
@@ -32,8 +32,10 @@
#include "xenstored_watch.h"
#include <xenctrl.h>
+#include <xen/grant_table.h>
static xc_interface **xc_handle;
+static xc_gnttab **xcg_handle;
static evtchn_port_t virq_port;
xc_evtchn *xce_handle = NULL;
@@ -163,6 +165,26 @@ static int readchn(struct connection *conn, void *data, unsigned int len)
return len;
}
+static void *map_interface(domid_t domid, unsigned long mfn)
+{
+ if (*xcg_handle >= 0) {
+ /* this is the preferred method */
+ return xc_gnttab_map_grant_ref(*xcg_handle, domid,
+ GNTTAB_RESERVED_XENSTORE, PROT_READ|PROT_WRITE);
+ } else {
+ return xc_map_foreign_range(*xc_handle, domid,
+ getpagesize(), PROT_READ|PROT_WRITE, mfn);
+ }
+}
+
+static void unmap_interface(void *interface)
+{
+ if (*xcg_handle >= 0)
+ xc_gnttab_munmap(*xcg_handle, interface, 1);
+ else
+ munmap(interface, getpagesize());
+}
+
static int destroy_domain(void *_domain)
{
struct domain *domain = _domain;
@@ -174,8 +196,12 @@ static int destroy_domain(void *_domain)
eprintf("> Unbinding port %i failed!\n", domain->port);
}
- if (domain->interface)
- munmap(domain->interface, getpagesize());
+ if (domain->interface) {
+ if (domain->domid == 0)
+ munmap(domain->interface, getpagesize());
+ else
+ unmap_interface(domain->interface);
+ }
fire_watches(NULL, "@releaseDomain", false);
@@ -344,9 +370,7 @@ void do_introduce(struct connection *conn, struct buffered_data *in)
domain = find_domain_by_domid(domid);
if (domain == NULL) {
- interface = xc_map_foreign_range(
- *xc_handle, domid,
- getpagesize(), PROT_READ|PROT_WRITE, mfn);
+ interface = map_interface(domid, mfn);
if (!interface) {
send_error(conn, errno);
return;
@@ -354,7 +378,7 @@ void do_introduce(struct connection *conn, struct buffered_data *in)
/* Hang domain off "in" until we're finished. */
domain = new_domain(in, domid, port);
if (!domain) {
- munmap(interface, getpagesize());
+ unmap_interface(interface);
send_error(conn, errno);
return;
}
@@ -552,6 +576,12 @@ static int close_xc_handle(void *_handle)
return 0;
}
+static int close_xcg_handle(void *_handle)
+{
+ xc_gnttab_close(*(xc_gnttab **)_handle);
+ return 0;
+}
+
/* Returns the implicit path of a connection (only domains have this) */
const char *get_implicit_path(const struct connection *conn)
{
@@ -603,6 +633,16 @@ void domain_init(void)
talloc_set_destructor(xc_handle, close_xc_handle);
+ xcg_handle = talloc(talloc_autofree_context(), xc_gnttab*);
+ if (!xcg_handle)
+ barf_perror("Failed to allocate domain gnttab handle");
+
+ *xcg_handle = xc_gnttab_open(NULL, 0);
+ if (*xcg_handle < 0)
+ xprintf("WARNING: Failed to open connection to gnttab\n");
+ else
+ talloc_set_destructor(xcg_handle, close_xcg_handle);
+
xce_handle = xc_evtchn_open(NULL, 0);
if (xce_handle == NULL)
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 15/23] xenstored: refactor socket setup code
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (13 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 14/23] xenstored: use grant references instead of map_foreign_range Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 16/23] xenstored: add NO_SOCKETS compilation option Daniel De Graaf
` (7 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/xenstore/xenstored_core.c | 85 +++++++++++++++++++++------------------
1 files changed, 46 insertions(+), 39 deletions(-)
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index 9e6c2c7..0e7d43f 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -343,13 +343,6 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
return max;
}
-static int destroy_fd(void *_fd)
-{
- int *fd = _fd;
- close(*fd);
- return 0;
-}
-
/* Is child a subnode of parent, or equal? */
bool is_child(const char *child, const char *parent)
{
@@ -1705,6 +1698,51 @@ static void daemonize(void)
umask(0);
}
+static int destroy_fd(void *_fd)
+{
+ int *fd = _fd;
+ close(*fd);
+ return 0;
+}
+
+static void init_sockets(int **psock, int **pro_sock)
+{
+ struct sockaddr_un addr;
+ int *sock, *ro_sock;
+ /* Create sockets for them to listen to. */
+ *psock = sock = talloc(talloc_autofree_context(), int);
+ *sock = socket(PF_UNIX, SOCK_STREAM, 0);
+ if (*sock < 0)
+ barf_perror("Could not create socket");
+ *pro_sock = ro_sock = talloc(talloc_autofree_context(), int);
+ *ro_sock = socket(PF_UNIX, SOCK_STREAM, 0);
+ if (*ro_sock < 0)
+ barf_perror("Could not create socket");
+ talloc_set_destructor(sock, destroy_fd);
+ talloc_set_destructor(ro_sock, destroy_fd);
+
+ /* FIXME: Be more sophisticated, don't mug running daemon. */
+ unlink(xs_daemon_socket());
+ unlink(xs_daemon_socket_ro());
+
+ addr.sun_family = AF_UNIX;
+ strcpy(addr.sun_path, xs_daemon_socket());
+ if (bind(*sock, (struct sockaddr *)&addr, sizeof(addr)) != 0)
+ barf_perror("Could not bind socket to %s", xs_daemon_socket());
+ strcpy(addr.sun_path, xs_daemon_socket_ro());
+ if (bind(*ro_sock, (struct sockaddr *)&addr, sizeof(addr)) != 0)
+ barf_perror("Could not bind socket to %s",
+ xs_daemon_socket_ro());
+ if (chmod(xs_daemon_socket(), 0600) != 0
+ || chmod(xs_daemon_socket_ro(), 0660) != 0)
+ barf_perror("Could not chmod sockets");
+
+ if (listen(*sock, 1) != 0
+ || listen(*ro_sock, 1) != 0)
+ barf_perror("Could not listen on sockets");
+
+
+}
static void usage(void)
{
@@ -1753,7 +1791,6 @@ extern void dump_conn(struct connection *conn);
int main(int argc, char *argv[])
{
int opt, *sock, *ro_sock, max;
- struct sockaddr_un addr;
fd_set inset, outset;
bool dofork = true;
bool outputpid = false;
@@ -1837,40 +1874,10 @@ int main(int argc, char *argv[])
if (!dofork)
talloc_enable_leak_report_full();
- /* Create sockets for them to listen to. */
- sock = talloc(talloc_autofree_context(), int);
- *sock = socket(PF_UNIX, SOCK_STREAM, 0);
- if (*sock < 0)
- barf_perror("Could not create socket");
- ro_sock = talloc(talloc_autofree_context(), int);
- *ro_sock = socket(PF_UNIX, SOCK_STREAM, 0);
- if (*ro_sock < 0)
- barf_perror("Could not create socket");
- talloc_set_destructor(sock, destroy_fd);
- talloc_set_destructor(ro_sock, destroy_fd);
-
/* Don't kill us with SIGPIPE. */
signal(SIGPIPE, SIG_IGN);
- /* FIXME: Be more sophisticated, don't mug running daemon. */
- unlink(xs_daemon_socket());
- unlink(xs_daemon_socket_ro());
-
- addr.sun_family = AF_UNIX;
- strcpy(addr.sun_path, xs_daemon_socket());
- if (bind(*sock, (struct sockaddr *)&addr, sizeof(addr)) != 0)
- barf_perror("Could not bind socket to %s", xs_daemon_socket());
- strcpy(addr.sun_path, xs_daemon_socket_ro());
- if (bind(*ro_sock, (struct sockaddr *)&addr, sizeof(addr)) != 0)
- barf_perror("Could not bind socket to %s",
- xs_daemon_socket_ro());
- if (chmod(xs_daemon_socket(), 0600) != 0
- || chmod(xs_daemon_socket_ro(), 0660) != 0)
- barf_perror("Could not chmod sockets");
-
- if (listen(*sock, 1) != 0
- || listen(*ro_sock, 1) != 0)
- barf_perror("Could not listen on sockets");
+ init_sockets(&sock, &ro_sock);
if (pipe(reopen_log_pipe)) {
barf_perror("pipe");
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 16/23] xenstored: add NO_SOCKETS compilation option
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (14 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 15/23] xenstored: refactor socket setup code Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 18:52 ` Stefano Stabellini
2012-01-25 14:37 ` [PATCH 17/23] xenstored: support for tdb_copy with TDB_INTERNAL Daniel De Graaf
` (6 subsequent siblings)
22 siblings, 1 reply; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Add option for compiling xenstored without unix sockets to support
running on mini-OS
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/xenstore/xenstored_core.c | 28 +++++++++++++++++++++++-----
1 files changed, 23 insertions(+), 5 deletions(-)
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index 0e7d43f..c1ee932 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -19,9 +19,11 @@
#include <sys/types.h>
#include <sys/stat.h>
-#include <sys/socket.h>
#include <sys/select.h>
+#ifndef NO_SOCKETS
+#include <sys/socket.h>
#include <sys/un.h>
+#endif
#include <sys/time.h>
#include <time.h>
#include <unistd.h>
@@ -320,8 +322,10 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
FD_ZERO(inset);
FD_ZERO(outset);
- set_fd(sock, inset, &max);
- set_fd(ro_sock, inset, &max);
+ if (sock != -1)
+ set_fd(sock, inset, &max);
+ if (ro_sock != -1)
+ set_fd(ro_sock, inset, &max);
set_fd(reopen_log_pipe[0], inset, &max);
if (xce_handle != NULL)
@@ -1345,6 +1349,11 @@ struct connection *new_connection(connwritefn_t *write, connreadfn_t *read)
return new;
}
+#ifdef NO_SOCKETS
+static void accept_connection(int sock, bool canwrite)
+{
+}
+#else
static int writefd(struct connection *conn, const void *data, unsigned int len)
{
int rc;
@@ -1399,6 +1408,7 @@ static void accept_connection(int sock, bool canwrite)
} else
close(fd);
}
+#endif
#define TDB_FLAGS 0
@@ -1698,6 +1708,13 @@ static void daemonize(void)
umask(0);
}
+#ifdef NO_SOCKETS
+static void init_sockets(int **psock, int **pro_sock)
+{
+ static int minus_one = -1;
+ *psock = *pro_sock = &minus_one;
+}
+#else
static int destroy_fd(void *_fd)
{
int *fd = _fd;
@@ -1743,6 +1760,7 @@ static void init_sockets(int **psock, int **pro_sock)
}
+#endif
static void usage(void)
{
@@ -1938,10 +1956,10 @@ int main(int argc, char *argv[])
reopen_log();
}
- if (FD_ISSET(*sock, &inset))
+ if (*sock != -1 && FD_ISSET(*sock, &inset))
accept_connection(*sock, true);
- if (FD_ISSET(*ro_sock, &inset))
+ if (*ro_sock != -1 && FD_ISSET(*ro_sock, &inset))
accept_connection(*ro_sock, false);
if (evtchn_fd != -1 && FD_ISSET(evtchn_fd, &inset))
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 17/23] xenstored: support for tdb_copy with TDB_INTERNAL
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (15 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 16/23] xenstored: add NO_SOCKETS compilation option Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 18/23] xenstored: support running in minios stubdom Daniel De Graaf
` (5 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
The tdb_copy function should honor the TDB_INTERNAL flag for in-memory
databases; this is required to run in mini-os which does not use a
filesystem.
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/xenstore/tdb.c | 35 +++++++++++++++++++++++++++++++++++
1 files changed, 35 insertions(+), 0 deletions(-)
diff --git a/tools/xenstore/tdb.c b/tools/xenstore/tdb.c
index 63205e1..3ecd3fc 100644
--- a/tools/xenstore/tdb.c
+++ b/tools/xenstore/tdb.c
@@ -2103,6 +2103,41 @@ TDB_CONTEXT *tdb_copy(TDB_CONTEXT *tdb, const char *outfile)
int fd, saved_errno;
TDB_CONTEXT *copy;
+ if (tdb->flags & TDB_INTERNAL) {
+ struct tdb_header *copydb;
+
+ copy = talloc_zero(outfile, TDB_CONTEXT);
+ if (copy == NULL) {
+ errno = ENOMEM;
+ goto intfail;
+ }
+ memcpy(copy, tdb, sizeof(TDB_CONTEXT));
+
+ if (copy->name || copy->locked || copy->device || copy->inode) {
+ fprintf(stderr, "tdb_copy assumption(s) failed\n");
+ goto intfail;
+ }
+
+ copydb = talloc_zero_size(copy, copy->map_size);
+ if (copydb == NULL) {
+ errno = ENOMEM;
+ goto intfail;
+ }
+ memcpy(copydb, copy->map_ptr, copy->map_size);
+ copy->map_ptr = (char*) copydb;
+
+ if (tdb_brlock(tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0) == -1)
+ goto intfail;
+
+ copy->next = tdbs;
+ tdbs = copy;
+
+ return copy;
+intfail:
+ talloc_free(copy);
+ return NULL;
+ }
+
fd = open(outfile, O_TRUNC|O_CREAT|O_WRONLY, 0640);
if (fd < 0)
return NULL;
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 18/23] xenstored: support running in minios stubdom
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (16 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 17/23] xenstored: support for tdb_copy with TDB_INTERNAL Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 19:14 ` Stefano Stabellini
2012-01-25 14:37 ` [PATCH 19/23] stubdom: enable xenstored build Daniel De Graaf
` (4 subsequent siblings)
22 siblings, 1 reply; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf, Alex Zeffertt, Diego Ongaro
A previous versions of this patch has been sent to xen-devel. See
http://lists.xensource.com/archives/html/xen-devel/2009-03/msg01655.html
Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/xenstore/Makefile | 9 ++++++++-
tools/xenstore/tdb.c | 6 +++---
tools/xenstore/utils.h | 2 ++
tools/xenstore/xenstored_core.c | 29 ++++++++++++++++++++++++++---
tools/xenstore/xenstored_domain.c | 11 +++++++++++
5 files changed, 50 insertions(+), 7 deletions(-)
diff --git a/tools/xenstore/Makefile b/tools/xenstore/Makefile
index 4facb62..be892fd 100644
--- a/tools/xenstore/Makefile
+++ b/tools/xenstore/Makefile
@@ -28,6 +28,10 @@ endif
ALL_TARGETS = libxenstore.so libxenstore.a clients xs_tdb_dump xenstored
+ifdef CONFIG_STUBDOM
+CFLAGS += -DNO_SOCKETS=1
+endif
+
.PHONY: all
all: $(ALL_TARGETS)
@@ -45,10 +49,13 @@ xenstored_probes.o: xenstored_solaris.o
CFLAGS += -DHAVE_DTRACE=1
endif
-
+
xenstored: $(XENSTORED_OBJS)
$(CC) $(LDFLAGS) $^ $(LDLIBS_libxenctrl) $(SOCKET_LIBS) -o $@ $(APPEND_LDFLAGS)
+xenstored.a: $(XENSTORED_OBJS)
+ $(AR) cr $@ $^
+
$(CLIENTS): xenstore
ln -f xenstore $@
diff --git a/tools/xenstore/tdb.c b/tools/xenstore/tdb.c
index 3ecd3fc..cb66ea7 100644
--- a/tools/xenstore/tdb.c
+++ b/tools/xenstore/tdb.c
@@ -1334,7 +1334,7 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
/* Iterate through chain */
while( tlock->off) {
- tdb_off current;
+ tdb_off mycurrent;
if (rec_read(tdb, tlock->off, rec) == -1)
goto fail;
@@ -1352,10 +1352,10 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
}
/* Try to clean dead ones from old traverses */
- current = tlock->off;
+ mycurrent = tlock->off;
tlock->off = rec->next;
if (!tdb->read_only &&
- do_delete(tdb, current, rec) != 0)
+ do_delete(tdb, mycurrent, rec) != 0)
goto fail;
}
tdb_unlock(tdb, tlock->hash, F_WRLCK);
diff --git a/tools/xenstore/utils.h b/tools/xenstore/utils.h
index f378343..2effd17 100644
--- a/tools/xenstore/utils.h
+++ b/tools/xenstore/utils.h
@@ -19,7 +19,9 @@ static inline bool strends(const char *a, const char *b)
return streq(a + strlen(a) - strlen(b), b);
}
+#ifndef ARRAY_SIZE
#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
+#endif
void barf(const char *fmt, ...) __attribute__((noreturn));
void barf_perror(const char *fmt, ...) __attribute__((noreturn));
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index c1ee932..7564edd 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -92,8 +92,10 @@ TDB_CONTEXT *tdb_context(struct connection *conn)
bool replace_tdb(const char *newname, TDB_CONTEXT *newtdb)
{
+#ifndef __MINIOS__
if (rename(newname, xs_daemon_tdb()) != 0)
return false;
+#endif
tdb_close(tdb_ctx);
tdb_ctx = talloc_steal(talloc_autofree_context(), newtdb);
return true;
@@ -223,7 +225,6 @@ static void reopen_log(void)
}
}
-
static bool write_messages(struct connection *conn)
{
int ret;
@@ -326,7 +327,8 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
set_fd(sock, inset, &max);
if (ro_sock != -1)
set_fd(ro_sock, inset, &max);
- set_fd(reopen_log_pipe[0], inset, &max);
+ if (reopen_log_pipe[0] != -1)
+ set_fd(reopen_log_pipe[0], inset, &max);
if (xce_handle != NULL)
set_fd(xc_evtchn_fd(xce_handle), inset, &max);
@@ -1410,7 +1412,11 @@ static void accept_connection(int sock, bool canwrite)
}
#endif
+#ifdef __MINIOS__
+#define TDB_FLAGS TDB_INTERNAL|TDB_NOLOCK
+#else
#define TDB_FLAGS 0
+#endif
/* We create initial nodes manually. */
static void manual_node(const char *name, const char *child)
@@ -1435,7 +1441,11 @@ static void setup_structure(void)
{
char *tdbname;
tdbname = talloc_strdup(talloc_autofree_context(), xs_daemon_tdb());
+#ifdef __MINIOS__
+ tdb_ctx = NULL;
+#else
tdb_ctx = tdb_open(tdbname, 0, TDB_FLAGS, O_RDWR, 0);
+#endif
if (tdb_ctx) {
/* XXX When we make xenstored able to restart, this will have
@@ -1661,6 +1671,7 @@ static void corrupt(struct connection *conn, const char *fmt, ...)
}
+#ifndef __MINIOS__
static void write_pidfile(const char *pidfile)
{
char buf[100];
@@ -1707,6 +1718,7 @@ static void daemonize(void)
/* Discard our parent's old-fashioned umask prejudices. */
umask(0);
}
+#endif
#ifdef NO_SOCKETS
static void init_sockets(int **psock, int **pro_sock)
@@ -1866,6 +1878,7 @@ int main(int argc, char *argv[])
reopen_log();
+#ifndef __MINIOS__
/* make sure xenstored directory exists */
if (mkdir(xs_daemon_rundir(), 0755)) {
if (errno != EEXIST) {
@@ -1887,6 +1900,7 @@ int main(int argc, char *argv[])
}
if (pidfile)
write_pidfile(pidfile);
+#endif
/* Talloc leak reports go to stderr, which is closed if we fork. */
if (!dofork)
@@ -1897,9 +1911,14 @@ int main(int argc, char *argv[])
init_sockets(&sock, &ro_sock);
+#ifdef __MINIOS__
+ reopen_log_pipe[0] = -1;
+ reopen_log_pipe[1] = -1;
+#else
if (pipe(reopen_log_pipe)) {
barf_perror("pipe");
}
+#endif
/* Setup the database */
setup_structure();
@@ -1916,6 +1935,7 @@ int main(int argc, char *argv[])
fflush(stdout);
}
+#ifndef __MINIOS__
/* redirect to /dev/null now we're ready to accept connections */
if (dofork) {
int devnull = open("/dev/null", O_RDWR);
@@ -1927,6 +1947,7 @@ int main(int argc, char *argv[])
close(devnull);
xprintf = trace;
}
+#endif
signal(SIGHUP, trigger_reopen_log);
@@ -1936,8 +1957,10 @@ int main(int argc, char *argv[])
/* Get ready to listen to the tools. */
max = initialize_set(&inset, &outset, *sock, *ro_sock, &timeout);
+#ifndef __MINIOS__
/* Tell the kernel we're up and running. */
xenbus_notify_running();
+#endif
/* Main loop. */
for (;;) {
@@ -1949,7 +1972,7 @@ int main(int argc, char *argv[])
barf_perror("Select failed");
}
- if (FD_ISSET(reopen_log_pipe[0], &inset)) {
+ if (reopen_log_pipe[0] != -1 && FD_ISSET(reopen_log_pipe[0], &inset)) {
char c;
if (read(reopen_log_pipe[0], &c, 1) != 1)
barf_perror("read failed");
diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
index 661d955..435f76a 100644
--- a/tools/xenstore/xenstored_domain.c
+++ b/tools/xenstore/xenstored_domain.c
@@ -197,10 +197,14 @@ static int destroy_domain(void *_domain)
}
if (domain->interface) {
+#ifdef __MINIOS__
+ unmap_interface(domain->interface);
+#else
if (domain->domid == 0)
munmap(domain->interface, getpagesize());
else
unmap_interface(domain->interface);
+#endif
}
fire_watches(NULL, "@releaseDomain", false);
@@ -595,6 +599,12 @@ void restore_existing_connections(void)
{
}
+#ifdef __MINIOS__
+static int dom0_init(void)
+{
+ return 0;
+}
+#else
static int dom0_init(void)
{
evtchn_port_t port;
@@ -618,6 +628,7 @@ static int dom0_init(void)
return 0;
}
+#endif
void domain_init(void)
{
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 19/23] stubdom: enable xenstored build
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (17 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 18/23] xenstored: support running in minios stubdom Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 20/23] xenstored: add --event parameter for bootstrapping Daniel De Graaf
` (3 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
stubdom/Makefile | 29 ++++++++++++++++++++++++++---
stubdom/xenstore-minios.cfg | 9 +++++++++
2 files changed, 35 insertions(+), 3 deletions(-)
create mode 100644 stubdom/xenstore-minios.cfg
diff --git a/stubdom/Makefile b/stubdom/Makefile
index d4da2bb..5c02464 100644
--- a/stubdom/Makefile
+++ b/stubdom/Makefile
@@ -74,14 +74,14 @@ TARGET_CPPFLAGS += -I$(XEN_ROOT)/xen/include
TARGET_LDFLAGS += -nostdlib -L$(CROSS_PREFIX)/$(GNU_TARGET_ARCH)-xen-elf/lib
-TARGETS=ioemu c caml grub
+TARGETS=ioemu c caml grub xenstore
CROSS_MAKE := $(MAKE) DESTDIR=
.PHONY: all
all: build
ifeq ($(STUBDOM_SUPPORTED),1)
-build: genpath ioemu-stubdom c-stubdom pv-grub
+build: genpath ioemu-stubdom c-stubdom pv-grub xenstore-stubdom
else
build: genpath
endif
@@ -262,6 +262,11 @@ mk-headers-$(XEN_TARGET_ARCH): ioemu/linkfarm.stamp
ln -sf $(XEN_ROOT)/tools/libxc/$(XEN_TARGET_ARCH)/*.c . && \
ln -sf $(XEN_ROOT)/tools/libxc/$(XEN_TARGET_ARCH)/*.h . && \
ln -sf $(XEN_ROOT)/tools/libxc/$(XEN_TARGET_ARCH)/Makefile . )
+ mkdir -p xenstore
+ [ -h xenstore/Makefile ] || ( cd xenstore && \
+ ln -sf $(XEN_ROOT)/tools/xenstore/*.c . && \
+ ln -sf $(XEN_ROOT)/tools/xenstore/*.h . && \
+ ln -sf $(XEN_ROOT)/tools/xenstore/Makefile . )
$(CROSS_MAKE) -C $(MINI_OS) links
touch mk-headers-$(XEN_TARGET_ARCH)
@@ -334,6 +339,14 @@ grub: grub-upstream $(CROSS_ROOT)
mkdir -p grub-$(XEN_TARGET_ARCH)
CPPFLAGS="$(TARGET_CPPFLAGS)" CFLAGS="$(TARGET_CFLAGS)" $(CROSS_MAKE) -C $@ OBJ_DIR=$(CURDIR)/grub-$(XEN_TARGET_ARCH)
+##########
+# xenstore
+##########
+
+.PHONY: xenstore
+xenstore: $(CROSS_ROOT)
+ CPPFLAGS="$(TARGET_CPPFLAGS)" CFLAGS="$(TARGET_CFLAGS)" $(CROSS_MAKE) -C $@ LWIPDIR=$(CURDIR)/lwip xenstored.a CONFIG_STUBDOM=y
+
########
# minios
########
@@ -355,12 +368,16 @@ c-stubdom: mini-os-$(XEN_TARGET_ARCH)-c lwip-$(XEN_TARGET_ARCH) libxc c
pv-grub: mini-os-$(XEN_TARGET_ARCH)-grub libxc grub
DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/grub/minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< APP_OBJS=$(CURDIR)/grub-$(XEN_TARGET_ARCH)/main.a
+.PHONY: xenstore-stubdom
+xenstore-stubdom: mini-os-$(XEN_TARGET_ARCH)-xenstore libxc xenstore
+ DEF_CPPFLAGS="$(TARGET_CPPFLAGS)" DEF_CFLAGS="$(TARGET_CFLAGS)" DEF_LDFLAGS="$(TARGET_LDFLAGS)" MINIOS_CONFIG="$(CURDIR)/xenstore-minios.cfg" $(CROSS_MAKE) -C $(MINI_OS) OBJ_DIR=$(CURDIR)/$< LWIPDIR=$(CURDIR)/lwip-$(XEN_TARGET_ARCH) APP_OBJS=$(CURDIR)/xenstore/xenstored.a
+
#########
# install
#########
ifeq ($(STUBDOM_SUPPORTED),1)
-install: genpath install-readme install-ioemu install-grub
+install: genpath install-readme install-ioemu install-grub install-xenstore
else
install: genpath
endif
@@ -379,6 +396,10 @@ install-grub: pv-grub
$(INSTALL_DIR) "$(DESTDIR)$(XENFIRMWAREDIR)"
$(INSTALL_DATA) mini-os-$(XEN_TARGET_ARCH)-grub/mini-os.gz "$(DESTDIR)$(XENFIRMWAREDIR)/pv-grub-$(XEN_TARGET_ARCH).gz"
+install-xenstore: xenstore-stubdom
+ $(INSTALL_DIR) "$(DESTDIR)/usr/lib/xen/boot"
+ $(INSTALL_PROG) mini-os-$(XEN_TARGET_ARCH)-xenstore/mini-os.gz "$(DESTDIR)/usr/lib/xen/boot/xenstore-stubdom.gz"
+
#######
# clean
#######
@@ -390,12 +411,14 @@ clean:
rm -fr mini-os-$(XEN_TARGET_ARCH)-c
rm -fr mini-os-$(XEN_TARGET_ARCH)-caml
rm -fr mini-os-$(XEN_TARGET_ARCH)-grub
+ rm -fr mini-os-$(XEN_TARGET_ARCH)-xenstore
$(CROSS_MAKE) -C caml clean
$(CROSS_MAKE) -C c clean
rm -fr grub-$(XEN_TARGET_ARCH)
rm -f $(STUBDOMPATH)
[ ! -d libxc-$(XEN_TARGET_ARCH) ] || $(CROSS_MAKE) -C libxc-$(XEN_TARGET_ARCH) clean
-[ ! -d ioemu ] || $(CROSS_MAKE) -C ioemu clean
+ -[ ! -d xenstore ] || $(CROSS_MAKE) -C xenstore clean
# clean the cross-compilation result
.PHONY: crossclean
diff --git a/stubdom/xenstore-minios.cfg b/stubdom/xenstore-minios.cfg
new file mode 100644
index 0000000..ab7338f
--- /dev/null
+++ b/stubdom/xenstore-minios.cfg
@@ -0,0 +1,9 @@
+CONFIG_BLKFRONT=n
+CONFIG_NETFRONT=n
+CONFIG_FBFRONT=n
+CONFIG_KBDFRONT=n
+CONFIG_CONSFRONT=n
+CONFIG_XENBUS=n
+
+lwip=n
+DEF_CPPFLAGS := $(filter-out -DHAVE_LWIP,$(DEF_CPPFLAGS))
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 20/23] xenstored: add --event parameter for bootstrapping
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (18 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 19/23] stubdom: enable xenstored build Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 21/23] xenstored: use domain_is_unprivileged instead of checking conn->id Daniel De Graaf
` (2 subsequent siblings)
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
When xenstored is run in a minios domain, it needs a bootstrap
connection to dom0 so that additional domain introduce messages can be
sent to it.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/xenstore/xenstored_core.c | 5 +++++
tools/xenstore/xenstored_core.h | 1 +
tools/xenstore/xenstored_domain.c | 11 +++++++++++
3 files changed, 17 insertions(+), 0 deletions(-)
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index 7564edd..dee7bbd 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -1804,6 +1804,7 @@ static struct option options[] = {
{ "no-domain-init", 0, NULL, 'D' },
{ "entry-nb", 1, NULL, 'E' },
{ "pid-file", 1, NULL, 'F' },
+ { "event", 1, NULL, 'e' },
{ "help", 0, NULL, 'H' },
{ "no-fork", 0, NULL, 'N' },
{ "output-pid", 0, NULL, 'P' },
@@ -1817,6 +1818,7 @@ static struct option options[] = {
{ NULL, 0, NULL, 0 } };
extern void dump_conn(struct connection *conn);
+int dom0_event = 0;
int main(int argc, char *argv[])
{
@@ -1871,6 +1873,9 @@ int main(int argc, char *argv[])
case 'W':
quota_nb_watch_per_domain = strtol(optarg, NULL, 10);
break;
+ case 'e':
+ dom0_event = strtol(optarg, NULL, 10);
+ break;
}
}
if (optind != argc)
diff --git a/tools/xenstore/xenstored_core.h b/tools/xenstore/xenstored_core.h
index c487089..d3040ba 100644
--- a/tools/xenstore/xenstored_core.h
+++ b/tools/xenstore/xenstored_core.h
@@ -168,6 +168,7 @@ void trace(const char *fmt, ...);
void dtrace_io(const struct connection *conn, const struct buffered_data *data, int out);
extern int event_fd;
+extern int dom0_event;
/* Map the kernel's xenstore page. */
void *xenbus_map(void);
diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
index 435f76a..6a0dbc2 100644
--- a/tools/xenstore/xenstored_domain.c
+++ b/tools/xenstore/xenstored_domain.c
@@ -602,6 +602,17 @@ void restore_existing_connections(void)
#ifdef __MINIOS__
static int dom0_init(void)
{
+ struct domain *domain;
+ int domid = 0;
+ evtchn_port_t port = dom0_event;
+
+ domain = new_domain(NULL, domid, port);
+ domain->interface = xc_gnttab_map_grant_ref(*xcg_handle, domid,
+ GNTTAB_RESERVED_XENSTORE, PROT_READ|PROT_WRITE);
+ talloc_steal(domain->conn, domain);
+
+ xc_evtchn_notify(xce_handle, domain->port);
+
return 0;
}
#else
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 21/23] xenstored: use domain_is_unprivileged instead of checking conn->id
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (19 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 20/23] xenstored: add --event parameter for bootstrapping Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 22/23] xenstored: add --priv-domid parameter Daniel De Graaf
2012-01-25 14:37 ` [PATCH 23/23] xenstored: Add stub domain builder Daniel De Graaf
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
This centralizes all the permission checking for privileged domains in
preparation for allowing domains other than dom0 to be privileged.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/xenstore/xenstored_core.c | 6 +++---
tools/xenstore/xenstored_domain.c | 8 ++++----
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index dee7bbd..06703ec 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -463,7 +463,7 @@ static enum xs_perm_type perm_for_conn(struct connection *conn,
mask &= ~XS_PERM_WRITE;
/* Owners and tools get it all... */
- if (!conn->id || perms[0].id == conn->id
+ if (!domain_is_unprivileged(conn) || perms[0].id == conn->id
|| (conn->target && perms[0].id == conn->target->id))
return (XS_PERM_READ|XS_PERM_WRITE|XS_PERM_OWNER) & mask;
@@ -801,11 +801,11 @@ static struct node *construct_node(struct connection *conn, const char *name)
node->tdb = tdb_context(conn);
node->name = talloc_strdup(node, name);
- /* Inherit permissions, except domains own what they create */
+ /* Inherit permissions, except unprivileged domains own what they create */
node->num_perms = parent->num_perms;
node->perms = talloc_memdup(node, parent->perms,
node->num_perms * sizeof(node->perms[0]));
- if (conn && conn->id)
+ if (domain_is_unprivileged(conn))
node->perms[0].id = conn->id;
/* No children, no data */
diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
index 6a0dbc2..d89528f 100644
--- a/tools/xenstore/xenstored_domain.c
+++ b/tools/xenstore/xenstored_domain.c
@@ -356,7 +356,7 @@ void do_introduce(struct connection *conn, struct buffered_data *in)
return;
}
- if (conn->id != 0 || !conn->can_write) {
+ if (domain_is_unprivileged(conn) || !conn->can_write) {
send_error(conn, EACCES);
return;
}
@@ -420,7 +420,7 @@ void do_set_target(struct connection *conn, struct buffered_data *in)
return;
}
- if (conn->id != 0 || !conn->can_write) {
+ if (domain_is_unprivileged(conn) || !conn->can_write) {
send_error(conn, EACCES);
return;
}
@@ -472,7 +472,7 @@ void do_release(struct connection *conn, const char *domid_str)
return;
}
- if (conn->id != 0) {
+ if (domain_is_unprivileged(conn)) {
send_error(conn, EACCES);
return;
}
@@ -509,7 +509,7 @@ void do_resume(struct connection *conn, const char *domid_str)
return;
}
- if (conn->id != 0) {
+ if (domain_is_unprivileged(conn)) {
send_error(conn, EACCES);
return;
}
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 22/23] xenstored: add --priv-domid parameter
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (20 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 21/23] xenstored: use domain_is_unprivileged instead of checking conn->id Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 23/23] xenstored: Add stub domain builder Daniel De Graaf
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
This parameter identifies an alternative service domain which has
superuser access to the xenstore database, which is currently required
to set up a new domain's xenstore entries.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
tools/xenstore/xenstored_core.c | 5 +++++
tools/xenstore/xenstored_core.h | 1 +
tools/xenstore/xenstored_domain.c | 2 +-
3 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
index 06703ec..6e0f782 100644
--- a/tools/xenstore/xenstored_core.c
+++ b/tools/xenstore/xenstored_core.c
@@ -1807,6 +1807,7 @@ static struct option options[] = {
{ "event", 1, NULL, 'e' },
{ "help", 0, NULL, 'H' },
{ "no-fork", 0, NULL, 'N' },
+ { "priv-domid", 1, NULL, 'p' },
{ "output-pid", 0, NULL, 'P' },
{ "entry-size", 1, NULL, 'S' },
{ "trace-file", 1, NULL, 'T' },
@@ -1819,6 +1820,7 @@ static struct option options[] = {
extern void dump_conn(struct connection *conn);
int dom0_event = 0;
+int priv_domid = 0;
int main(int argc, char *argv[])
{
@@ -1876,6 +1878,9 @@ int main(int argc, char *argv[])
case 'e':
dom0_event = strtol(optarg, NULL, 10);
break;
+ case 'p':
+ priv_domid = strtol(optarg, NULL, 10);
+ break;
}
}
if (optind != argc)
diff --git a/tools/xenstore/xenstored_core.h b/tools/xenstore/xenstored_core.h
index d3040ba..03e2e48 100644
--- a/tools/xenstore/xenstored_core.h
+++ b/tools/xenstore/xenstored_core.h
@@ -169,6 +169,7 @@ void dtrace_io(const struct connection *conn, const struct buffered_data *data,
extern int event_fd;
extern int dom0_event;
+extern int priv_domid;
/* Map the kernel's xenstore page. */
void *xenbus_map(void);
diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
index d89528f..8c215fb 100644
--- a/tools/xenstore/xenstored_domain.c
+++ b/tools/xenstore/xenstored_domain.c
@@ -261,7 +261,7 @@ bool domain_can_read(struct connection *conn)
bool domain_is_unprivileged(struct connection *conn)
{
- return (conn && conn->domain && conn->domain->domid != 0);
+ return (conn && conn->domain && conn->domain->domid != 0 && conn->domain->domid != priv_domid);
}
bool domain_can_write(struct connection *conn)
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* [PATCH 23/23] xenstored: Add stub domain builder
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
` (21 preceding siblings ...)
2012-01-25 14:37 ` [PATCH 22/23] xenstored: add --priv-domid parameter Daniel De Graaf
@ 2012-01-25 14:37 ` Daniel De Graaf
22 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 14:37 UTC (permalink / raw)
To: xen-devel; +Cc: Daniel De Graaf
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
---
tools/include/xen-sys/Linux/xenbus_dev.h | 44 ++++++++++++++
tools/xenstore/Makefile | 9 ++-
tools/xenstore/init-xenstore-domain.c | 94 ++++++++++++++++++++++++++++++
3 files changed, 145 insertions(+), 2 deletions(-)
create mode 100644 tools/include/xen-sys/Linux/xenbus_dev.h
create mode 100644 tools/xenstore/init-xenstore-domain.c
diff --git a/tools/include/xen-sys/Linux/xenbus_dev.h b/tools/include/xen-sys/Linux/xenbus_dev.h
new file mode 100644
index 0000000..bbee8c6
--- /dev/null
+++ b/tools/include/xen-sys/Linux/xenbus_dev.h
@@ -0,0 +1,44 @@
+/******************************************************************************
+ * evtchn.h
+ *
+ * Interface to /dev/xen/xenbus_backend.
+ *
+ * Copyright (c) 2011 Bastian Blank <waldi@debian.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef __LINUX_XEN_XENBUS_DEV_H__
+#define __LINUX_XEN_XENBUS_DEV_H__
+
+#include <linux/ioctl.h>
+
+#define IOCTL_XENBUS_BACKEND_EVTCHN \
+ _IOC(_IOC_NONE, 'B', 0, 0)
+
+#define IOCTL_XENBUS_BACKEND_SETUP \
+ _IOC(_IOC_NONE, 'B', 1, 0)
+
+#endif /* __LINUX_XEN_XENBUS_DEV_H__ */
diff --git a/tools/xenstore/Makefile b/tools/xenstore/Makefile
index be892fd..2fce994 100644
--- a/tools/xenstore/Makefile
+++ b/tools/xenstore/Makefile
@@ -26,7 +26,7 @@ LIBXENSTORE := libxenstore.a
xenstore xenstore-control: CFLAGS += -static
endif
-ALL_TARGETS = libxenstore.so libxenstore.a clients xs_tdb_dump xenstored
+ALL_TARGETS = libxenstore.so libxenstore.a clients xs_tdb_dump xenstored init-xenstore-domain
ifdef CONFIG_STUBDOM
CFLAGS += -DNO_SOCKETS=1
@@ -50,6 +50,11 @@ xenstored_probes.o: xenstored_solaris.o
CFLAGS += -DHAVE_DTRACE=1
endif
+init-xenstore-domain.o: CFLAGS += $(CFLAGS_libxenguest)
+
+init-xenstore-domain: init-xenstore-domain.o $(LIBXENSTORE)
+ $(CC) $(LDFLAGS) $^ $(LDLIBS_libxenctrl) $(LDLIBS_libxenguest) $(LDLIBS_libxenstore) -o $@ $(APPEND_LDFLAGS)
+
xenstored: $(XENSTORED_OBJS)
$(CC) $(LDFLAGS) $^ $(LDLIBS_libxenctrl) $(SOCKET_LIBS) -o $@ $(APPEND_LDFLAGS)
@@ -85,7 +90,7 @@ libxenstore.a: xs.o xs_lib.o
clean:
rm -f *.a *.o *.opic *.so* xenstored_probes.h
rm -f xenstored xs_random xs_stress xs_crashme
- rm -f xs_tdb_dump xenstore-control
+ rm -f xs_tdb_dump xenstore-control init-xenstore-domain
rm -f xenstore $(CLIENTS)
$(RM) $(DEPS)
diff --git a/tools/xenstore/init-xenstore-domain.c b/tools/xenstore/init-xenstore-domain.c
new file mode 100644
index 0000000..f6c31d0
--- /dev/null
+++ b/tools/xenstore/init-xenstore-domain.c
@@ -0,0 +1,94 @@
+#include <fcntl.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <xenctrl.h>
+#include <xc_dom.h>
+#include <xs.h>
+#include <xen/sys/xenbus_dev.h>
+
+static uint32_t domid = -1;
+
+static int build(xc_interface *xch, char** argv)
+{
+ char cmdline[512];
+ uint32_t ssid;
+ xen_domain_handle_t handle = { 0 };
+ int rv;
+ int xs_fd = open("/dev/xen/xenbus_backend", O_RDWR);
+ struct xc_dom_image *dom;
+ int maxmem = atoi(argv[2]);
+ int limit_kb = (maxmem + 1)*1024;
+
+ rv = xc_flask_context_to_sid(xch, argv[3], strlen(argv[3]), &ssid);
+ if (rv) return rv;
+ rv = xc_domain_create(xch, ssid, handle, 0, &domid);
+ if (rv) return rv;
+ rv = xc_domain_max_vcpus(xch, domid, 1);
+ if (rv) return rv;
+ rv = xc_domain_setmaxmem(xch, domid, limit_kb);
+ if (rv) return rv;
+ rv = xc_domain_set_memmap_limit(xch, domid, limit_kb);
+ if (rv) return rv;
+
+ rv = ioctl(xs_fd, IOCTL_XENBUS_BACKEND_SETUP, domid);
+ if (rv < 0) return rv;
+ snprintf(cmdline, 512, "--event %d", rv);
+
+ dom = xc_dom_allocate(xch, cmdline, NULL);
+ rv = xc_dom_kernel_file(dom, argv[1]);
+ if (rv) return rv;
+ rv = xc_dom_boot_xen_init(dom, xch, domid);
+ if (rv) return rv;
+ rv = xc_dom_parse_image(dom);
+ if (rv) return rv;
+ rv = xc_dom_mem_init(dom, maxmem);
+ if (rv) return rv;
+ rv = xc_dom_boot_mem_init(dom);
+ if (rv) return rv;
+ rv = xc_dom_build_image(dom);
+ if (rv) return rv;
+ rv = xc_dom_boot_image(dom);
+ if (rv) return rv;
+
+ xc_dom_release(dom);
+
+ rv = xc_domain_set_virq_handler(xch, domid, VIRQ_DOM_EXC);
+ if (rv) return rv;
+ rv = xc_domain_unpause(xch, domid);
+ if (rv) return rv;
+
+ return 0;
+}
+
+int main(int argc, char** argv)
+{
+ xc_interface *xch;
+ struct xs_handle *xsh;
+ char buf[16];
+ int rv;
+
+ if (argc != 4) {
+ printf("Use: %s <xenstore-kernel> <memory_mb> <flask-label>\n", argv[0]);
+ return 2;
+ }
+
+ xch = xc_interface_open(NULL, NULL, 0);
+ if (!xch) return 1;
+
+ rv = build(xch, argv);
+
+ xc_interface_close(xch);
+
+ if (rv) return 1;
+
+ xsh = xs_open(0);
+ rv = snprintf(buf, 16, "%d", domid);
+ xs_write(xsh, XBT_NULL, "/tool/xenstored/domid", buf, rv);
+ xs_daemon_close(xsh);
+
+ return 0;
+}
--
1.7.7.5
^ permalink raw reply related [flat|nested] 27+ messages in thread
* Re: [PATCH 16/23] xenstored: add NO_SOCKETS compilation option
2012-01-25 14:37 ` [PATCH 16/23] xenstored: add NO_SOCKETS compilation option Daniel De Graaf
@ 2012-01-25 18:52 ` Stefano Stabellini
0 siblings, 0 replies; 27+ messages in thread
From: Stefano Stabellini @ 2012-01-25 18:52 UTC (permalink / raw)
To: Daniel De Graaf; +Cc: Alex Zeffertt, xen-devel, Diego Ongaro
On Wed, 25 Jan 2012, Daniel De Graaf wrote:
> From: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
>
> Add option for compiling xenstored without unix sockets to support
> running on mini-OS
It looks much nicer now, thanks.
> Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
> Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
> Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
> ---
> tools/xenstore/xenstored_core.c | 28 +++++++++++++++++++++++-----
> 1 files changed, 23 insertions(+), 5 deletions(-)
>
> diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
> index 0e7d43f..c1ee932 100644
> --- a/tools/xenstore/xenstored_core.c
> +++ b/tools/xenstore/xenstored_core.c
> @@ -19,9 +19,11 @@
>
> #include <sys/types.h>
> #include <sys/stat.h>
> -#include <sys/socket.h>
> #include <sys/select.h>
> +#ifndef NO_SOCKETS
> +#include <sys/socket.h>
> #include <sys/un.h>
> +#endif
> #include <sys/time.h>
> #include <time.h>
> #include <unistd.h>
> @@ -320,8 +322,10 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
> FD_ZERO(inset);
> FD_ZERO(outset);
>
> - set_fd(sock, inset, &max);
> - set_fd(ro_sock, inset, &max);
> + if (sock != -1)
> + set_fd(sock, inset, &max);
> + if (ro_sock != -1)
> + set_fd(ro_sock, inset, &max);
> set_fd(reopen_log_pipe[0], inset, &max);
>
> if (xce_handle != NULL)
> @@ -1345,6 +1349,11 @@ struct connection *new_connection(connwritefn_t *write, connreadfn_t *read)
> return new;
> }
>
> +#ifdef NO_SOCKETS
> +static void accept_connection(int sock, bool canwrite)
> +{
> +}
> +#else
> static int writefd(struct connection *conn, const void *data, unsigned int len)
> {
> int rc;
> @@ -1399,6 +1408,7 @@ static void accept_connection(int sock, bool canwrite)
> } else
> close(fd);
> }
> +#endif
>
> #define TDB_FLAGS 0
>
> @@ -1698,6 +1708,13 @@ static void daemonize(void)
> umask(0);
> }
>
> +#ifdef NO_SOCKETS
> +static void init_sockets(int **psock, int **pro_sock)
> +{
> + static int minus_one = -1;
> + *psock = *pro_sock = &minus_one;
> +}
> +#else
> static int destroy_fd(void *_fd)
> {
> int *fd = _fd;
> @@ -1743,6 +1760,7 @@ static void init_sockets(int **psock, int **pro_sock)
>
>
> }
> +#endif
>
> static void usage(void)
> {
> @@ -1938,10 +1956,10 @@ int main(int argc, char *argv[])
> reopen_log();
> }
>
> - if (FD_ISSET(*sock, &inset))
> + if (*sock != -1 && FD_ISSET(*sock, &inset))
> accept_connection(*sock, true);
>
> - if (FD_ISSET(*ro_sock, &inset))
> + if (*ro_sock != -1 && FD_ISSET(*ro_sock, &inset))
> accept_connection(*ro_sock, false);
>
> if (evtchn_fd != -1 && FD_ISSET(evtchn_fd, &inset))
> --
> 1.7.7.5
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
^ permalink raw reply [flat|nested] 27+ messages in thread
* Re: [PATCH 18/23] xenstored: support running in minios stubdom
2012-01-25 14:37 ` [PATCH 18/23] xenstored: support running in minios stubdom Daniel De Graaf
@ 2012-01-25 19:14 ` Stefano Stabellini
2012-01-25 20:22 ` Daniel De Graaf
0 siblings, 1 reply; 27+ messages in thread
From: Stefano Stabellini @ 2012-01-25 19:14 UTC (permalink / raw)
To: Daniel De Graaf; +Cc: Alex Zeffertt, xen-devel, Diego Ongaro
On Wed, 25 Jan 2012, Daniel De Graaf wrote:
> A previous versions of this patch has been sent to xen-devel. See
> http://lists.xensource.com/archives/html/xen-devel/2009-03/msg01655.html
>
> Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
> Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
> Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
> ---
> tools/xenstore/Makefile | 9 ++++++++-
> tools/xenstore/tdb.c | 6 +++---
> tools/xenstore/utils.h | 2 ++
> tools/xenstore/xenstored_core.c | 29 ++++++++++++++++++++++++++---
> tools/xenstore/xenstored_domain.c | 11 +++++++++++
> 5 files changed, 50 insertions(+), 7 deletions(-)
>
> diff --git a/tools/xenstore/Makefile b/tools/xenstore/Makefile
> index 4facb62..be892fd 100644
> --- a/tools/xenstore/Makefile
> +++ b/tools/xenstore/Makefile
> @@ -28,6 +28,10 @@ endif
>
> ALL_TARGETS = libxenstore.so libxenstore.a clients xs_tdb_dump xenstored
>
> +ifdef CONFIG_STUBDOM
> +CFLAGS += -DNO_SOCKETS=1
> +endif
> +
> .PHONY: all
> all: $(ALL_TARGETS)
>
> @@ -45,10 +49,13 @@ xenstored_probes.o: xenstored_solaris.o
>
> CFLAGS += -DHAVE_DTRACE=1
> endif
> -
> +
> xenstored: $(XENSTORED_OBJS)
> $(CC) $(LDFLAGS) $^ $(LDLIBS_libxenctrl) $(SOCKET_LIBS) -o $@ $(APPEND_LDFLAGS)
>
> +xenstored.a: $(XENSTORED_OBJS)
> + $(AR) cr $@ $^
> +
> $(CLIENTS): xenstore
> ln -f xenstore $@
>
> diff --git a/tools/xenstore/tdb.c b/tools/xenstore/tdb.c
> index 3ecd3fc..cb66ea7 100644
> --- a/tools/xenstore/tdb.c
> +++ b/tools/xenstore/tdb.c
> @@ -1334,7 +1334,7 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
>
> /* Iterate through chain */
> while( tlock->off) {
> - tdb_off current;
> + tdb_off mycurrent;
> if (rec_read(tdb, tlock->off, rec) == -1)
> goto fail;
>
> @@ -1352,10 +1352,10 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
> }
>
> /* Try to clean dead ones from old traverses */
> - current = tlock->off;
> + mycurrent = tlock->off;
> tlock->off = rec->next;
> if (!tdb->read_only &&
> - do_delete(tdb, current, rec) != 0)
> + do_delete(tdb, mycurrent, rec) != 0)
> goto fail;
> }
> tdb_unlock(tdb, tlock->hash, F_WRLCK);
Why are you renaming current to mycurrent?
> diff --git a/tools/xenstore/utils.h b/tools/xenstore/utils.h
> index f378343..2effd17 100644
> --- a/tools/xenstore/utils.h
> +++ b/tools/xenstore/utils.h
> @@ -19,7 +19,9 @@ static inline bool strends(const char *a, const char *b)
> return streq(a + strlen(a) - strlen(b), b);
> }
>
> +#ifndef ARRAY_SIZE
> #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
> +#endif
>
> void barf(const char *fmt, ...) __attribute__((noreturn));
> void barf_perror(const char *fmt, ...) __attribute__((noreturn));
> diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
> index c1ee932..7564edd 100644
> --- a/tools/xenstore/xenstored_core.c
> +++ b/tools/xenstore/xenstored_core.c
> @@ -92,8 +92,10 @@ TDB_CONTEXT *tdb_context(struct connection *conn)
>
> bool replace_tdb(const char *newname, TDB_CONTEXT *newtdb)
> {
> +#ifndef __MINIOS__
> if (rename(newname, xs_daemon_tdb()) != 0)
> return false;
> +#endif
> tdb_close(tdb_ctx);
> tdb_ctx = talloc_steal(talloc_autofree_context(), newtdb);
> return true;
Are you doing this because the database is not on file anymore, correct?
Maybe there is some room for introducing this concept more generally
because it could be useful on Linux too and it would avoid this and
other ifdef's.
> @@ -223,7 +225,6 @@ static void reopen_log(void)
> }
> }
>
> -
> static bool write_messages(struct connection *conn)
> {
> int ret;
> @@ -326,7 +327,8 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
> set_fd(sock, inset, &max);
> if (ro_sock != -1)
> set_fd(ro_sock, inset, &max);
> - set_fd(reopen_log_pipe[0], inset, &max);
> + if (reopen_log_pipe[0] != -1)
> + set_fd(reopen_log_pipe[0], inset, &max);
>
> if (xce_handle != NULL)
> set_fd(xc_evtchn_fd(xce_handle), inset, &max);
> @@ -1410,7 +1412,11 @@ static void accept_connection(int sock, bool canwrite)
> }
> #endif
>
> +#ifdef __MINIOS__
> +#define TDB_FLAGS TDB_INTERNAL|TDB_NOLOCK
> +#else
> #define TDB_FLAGS 0
> +#endif
See above
> /* We create initial nodes manually. */
> static void manual_node(const char *name, const char *child)
> @@ -1435,7 +1441,11 @@ static void setup_structure(void)
> {
> char *tdbname;
> tdbname = talloc_strdup(talloc_autofree_context(), xs_daemon_tdb());
> +#ifdef __MINIOS__
> + tdb_ctx = NULL;
> +#else
> tdb_ctx = tdb_open(tdbname, 0, TDB_FLAGS, O_RDWR, 0);
> +#endif
ditto
> if (tdb_ctx) {
> /* XXX When we make xenstored able to restart, this will have
> @@ -1661,6 +1671,7 @@ static void corrupt(struct connection *conn, const char *fmt, ...)
> }
>
>
> +#ifndef __MINIOS__
> static void write_pidfile(const char *pidfile)
> {
> char buf[100];
> @@ -1707,6 +1718,7 @@ static void daemonize(void)
> /* Discard our parent's old-fashioned umask prejudices. */
> umask(0);
> }
> +#endif
>
> #ifdef NO_SOCKETS
> static void init_sockets(int **psock, int **pro_sock)
> @@ -1866,6 +1878,7 @@ int main(int argc, char *argv[])
>
> reopen_log();
>
> +#ifndef __MINIOS__
> /* make sure xenstored directory exists */
> if (mkdir(xs_daemon_rundir(), 0755)) {
> if (errno != EEXIST) {
> @@ -1887,6 +1900,7 @@ int main(int argc, char *argv[])
> }
> if (pidfile)
> write_pidfile(pidfile);
> +#endif
Maybe we could just have this fail gracefully: after all even on Linux
if the socket interface is not available everything should be able to
run correctly.
> /* Talloc leak reports go to stderr, which is closed if we fork. */
> if (!dofork)
> @@ -1897,9 +1911,14 @@ int main(int argc, char *argv[])
>
> init_sockets(&sock, &ro_sock);
>
> +#ifdef __MINIOS__
> + reopen_log_pipe[0] = -1;
> + reopen_log_pipe[1] = -1;
> +#else
> if (pipe(reopen_log_pipe)) {
> barf_perror("pipe");
> }
> +#endif
>
> /* Setup the database */
> setup_structure();
> @@ -1916,6 +1935,7 @@ int main(int argc, char *argv[])
> fflush(stdout);
> }
>
> +#ifndef __MINIOS__
> /* redirect to /dev/null now we're ready to accept connections */
> if (dofork) {
> int devnull = open("/dev/null", O_RDWR);
> @@ -1927,6 +1947,7 @@ int main(int argc, char *argv[])
> close(devnull);
> xprintf = trace;
> }
> +#endif
we could probably handle /dev/null in minios from sys.c
> signal(SIGHUP, trigger_reopen_log);
>
> @@ -1936,8 +1957,10 @@ int main(int argc, char *argv[])
> /* Get ready to listen to the tools. */
> max = initialize_set(&inset, &outset, *sock, *ro_sock, &timeout);
>
> +#ifndef __MINIOS__
> /* Tell the kernel we're up and running. */
> xenbus_notify_running();
> +#endif
>
> /* Main loop. */
> for (;;) {
> @@ -1949,7 +1972,7 @@ int main(int argc, char *argv[])
> barf_perror("Select failed");
> }
>
> - if (FD_ISSET(reopen_log_pipe[0], &inset)) {
> + if (reopen_log_pipe[0] != -1 && FD_ISSET(reopen_log_pipe[0], &inset)) {
> char c;
> if (read(reopen_log_pipe[0], &c, 1) != 1)
> barf_perror("read failed");
> diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
> index 661d955..435f76a 100644
> --- a/tools/xenstore/xenstored_domain.c
> +++ b/tools/xenstore/xenstored_domain.c
> @@ -197,10 +197,14 @@ static int destroy_domain(void *_domain)
> }
>
> if (domain->interface) {
> +#ifdef __MINIOS__
> + unmap_interface(domain->interface);
> +#else
> if (domain->domid == 0)
> munmap(domain->interface, getpagesize());
> else
> unmap_interface(domain->interface);
> +#endif
> }
Argh.
Why dom0 cannot be handled inside unmap_interface?
> fire_watches(NULL, "@releaseDomain", false);
> @@ -595,6 +599,12 @@ void restore_existing_connections(void)
> {
> }
>
> +#ifdef __MINIOS__
> +static int dom0_init(void)
> +{
> + return 0;
> +}
> +#else
> static int dom0_init(void)
> {
> evtchn_port_t port;
> @@ -618,6 +628,7 @@ static int dom0_init(void)
>
> return 0;
> }
> +#endif
>
> void domain_init(void)
> {
^ permalink raw reply [flat|nested] 27+ messages in thread
* Re: [PATCH 18/23] xenstored: support running in minios stubdom
2012-01-25 19:14 ` Stefano Stabellini
@ 2012-01-25 20:22 ` Daniel De Graaf
0 siblings, 0 replies; 27+ messages in thread
From: Daniel De Graaf @ 2012-01-25 20:22 UTC (permalink / raw)
To: Stefano Stabellini; +Cc: xen-devel
On 01/25/2012 02:14 PM, Stefano Stabellini wrote:
> On Wed, 25 Jan 2012, Daniel De Graaf wrote:
>> A previous versions of this patch has been sent to xen-devel. See
>> http://lists.xensource.com/archives/html/xen-devel/2009-03/msg01655.html
>>
>> Signed-off-by: Diego Ongaro <diego.ongaro@citrix.com>
>> Signed-off-by: Alex Zeffertt <alex.zeffertt@eu.citrix.com>
>> Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
>> ---
>> tools/xenstore/Makefile | 9 ++++++++-
>> tools/xenstore/tdb.c | 6 +++---
>> tools/xenstore/utils.h | 2 ++
>> tools/xenstore/xenstored_core.c | 29 ++++++++++++++++++++++++++---
>> tools/xenstore/xenstored_domain.c | 11 +++++++++++
>> 5 files changed, 50 insertions(+), 7 deletions(-)
>>
>> diff --git a/tools/xenstore/Makefile b/tools/xenstore/Makefile
>> index 4facb62..be892fd 100644
>> --- a/tools/xenstore/Makefile
>> +++ b/tools/xenstore/Makefile
>> @@ -28,6 +28,10 @@ endif
>>
>> ALL_TARGETS = libxenstore.so libxenstore.a clients xs_tdb_dump xenstored
>>
>> +ifdef CONFIG_STUBDOM
>> +CFLAGS += -DNO_SOCKETS=1
>> +endif
>> +
>> .PHONY: all
>> all: $(ALL_TARGETS)
>>
>> @@ -45,10 +49,13 @@ xenstored_probes.o: xenstored_solaris.o
>>
>> CFLAGS += -DHAVE_DTRACE=1
>> endif
>> -
>> +
>> xenstored: $(XENSTORED_OBJS)
>> $(CC) $(LDFLAGS) $^ $(LDLIBS_libxenctrl) $(SOCKET_LIBS) -o $@ $(APPEND_LDFLAGS)
>>
>> +xenstored.a: $(XENSTORED_OBJS)
>> + $(AR) cr $@ $^
>> +
>> $(CLIENTS): xenstore
>> ln -f xenstore $@
>>
>> diff --git a/tools/xenstore/tdb.c b/tools/xenstore/tdb.c
>> index 3ecd3fc..cb66ea7 100644
>> --- a/tools/xenstore/tdb.c
>> +++ b/tools/xenstore/tdb.c
>> @@ -1334,7 +1334,7 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
>>
>> /* Iterate through chain */
>> while( tlock->off) {
>> - tdb_off current;
>> + tdb_off mycurrent;
>> if (rec_read(tdb, tlock->off, rec) == -1)
>> goto fail;
>>
>> @@ -1352,10 +1352,10 @@ static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
>> }
>>
>> /* Try to clean dead ones from old traverses */
>> - current = tlock->off;
>> + mycurrent = tlock->off;
>> tlock->off = rec->next;
>> if (!tdb->read_only &&
>> - do_delete(tdb, current, rec) != 0)
>> + do_delete(tdb, mycurrent, rec) != 0)
>> goto fail;
>> }
>> tdb_unlock(tdb, tlock->hash, F_WRLCK);
>
> Why are you renaming current to mycurrent?
Hmm. While current is a #define in mini-os sched.h, this file doesn't
include that header so it shouldn't break. I'll drop this change.
>
>> diff --git a/tools/xenstore/utils.h b/tools/xenstore/utils.h
>> index f378343..2effd17 100644
>> --- a/tools/xenstore/utils.h
>> +++ b/tools/xenstore/utils.h
>> @@ -19,7 +19,9 @@ static inline bool strends(const char *a, const char *b)
>> return streq(a + strlen(a) - strlen(b), b);
>> }
>>
>> +#ifndef ARRAY_SIZE
>> #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
>> +#endif
>>
>> void barf(const char *fmt, ...) __attribute__((noreturn));
>> void barf_perror(const char *fmt, ...) __attribute__((noreturn));
>> diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c
>> index c1ee932..7564edd 100644
>> --- a/tools/xenstore/xenstored_core.c
>> +++ b/tools/xenstore/xenstored_core.c
>> @@ -92,8 +92,10 @@ TDB_CONTEXT *tdb_context(struct connection *conn)
>>
>> bool replace_tdb(const char *newname, TDB_CONTEXT *newtdb)
>> {
>> +#ifndef __MINIOS__
>> if (rename(newname, xs_daemon_tdb()) != 0)
>> return false;
>> +#endif
>> tdb_close(tdb_ctx);
>> tdb_ctx = talloc_steal(talloc_autofree_context(), newtdb);
>> return true;
>
> Are you doing this because the database is not on file anymore, correct?
> Maybe there is some room for introducing this concept more generally
> because it could be useful on Linux too and it would avoid this and
> other ifdef's.
Yes, this one can check TDB_INTERNAL (not sure why the original author
didn't want to check that flag; it works fine).
>
>> @@ -223,7 +225,6 @@ static void reopen_log(void)
>> }
>> }
>>
>> -
>> static bool write_messages(struct connection *conn)
>> {
>> int ret;
>> @@ -326,7 +327,8 @@ static int initialize_set(fd_set *inset, fd_set *outset, int sock, int ro_sock,
>> set_fd(sock, inset, &max);
>> if (ro_sock != -1)
>> set_fd(ro_sock, inset, &max);
>> - set_fd(reopen_log_pipe[0], inset, &max);
>> + if (reopen_log_pipe[0] != -1)
>> + set_fd(reopen_log_pipe[0], inset, &max);
>>
>> if (xce_handle != NULL)
>> set_fd(xc_evtchn_fd(xce_handle), inset, &max);
>> @@ -1410,7 +1412,11 @@ static void accept_connection(int sock, bool canwrite)
>> }
>> #endif
>>
>> +#ifdef __MINIOS__
>> +#define TDB_FLAGS TDB_INTERNAL|TDB_NOLOCK
>> +#else
>> #define TDB_FLAGS 0
>> +#endif
>
> See above
I will be adding a command-line argument for this, and force it to be
enabled if __MINIOS__.
>
>> /* We create initial nodes manually. */
>> static void manual_node(const char *name, const char *child)
>> @@ -1435,7 +1441,11 @@ static void setup_structure(void)
>> {
>> char *tdbname;
>> tdbname = talloc_strdup(talloc_autofree_context(), xs_daemon_tdb());
>> +#ifdef __MINIOS__
>> + tdb_ctx = NULL;
>> +#else
>> tdb_ctx = tdb_open(tdbname, 0, TDB_FLAGS, O_RDWR, 0);
>> +#endif
>
> ditto
>
>> if (tdb_ctx) {
>> /* XXX When we make xenstored able to restart, this will have
>> @@ -1661,6 +1671,7 @@ static void corrupt(struct connection *conn, const char *fmt, ...)
>> }
>>
>>
>> +#ifndef __MINIOS__
>> static void write_pidfile(const char *pidfile)
>> {
>> char buf[100];
>> @@ -1707,6 +1718,7 @@ static void daemonize(void)
>> /* Discard our parent's old-fashioned umask prejudices. */
>> umask(0);
>> }
>> +#endif
>>
>> #ifdef NO_SOCKETS
>> static void init_sockets(int **psock, int **pro_sock)
>> @@ -1866,6 +1878,7 @@ int main(int argc, char *argv[])
>>
>> reopen_log();
>>
>> +#ifndef __MINIOS__
>> /* make sure xenstored directory exists */
>> if (mkdir(xs_daemon_rundir(), 0755)) {
>> if (errno != EEXIST) {
>> @@ -1887,6 +1900,7 @@ int main(int argc, char *argv[])
>> }
>> if (pidfile)
>> write_pidfile(pidfile);
>> +#endif
>
> Maybe we could just have this fail gracefully: after all even on Linux
> if the socket interface is not available everything should be able to
> run correctly.
>
That's simple enough to change, although it might make debugging a bit harder
on Linux if there are stale socket files around with bad permissions.
Actually... this is just making the directories. Error reporting doesn't belong
here, it belongs where we open the files that go in the directories.
>
>> /* Talloc leak reports go to stderr, which is closed if we fork. */
>> if (!dofork)
>> @@ -1897,9 +1911,14 @@ int main(int argc, char *argv[])
>>
>> init_sockets(&sock, &ro_sock);
>>
>> +#ifdef __MINIOS__
>> + reopen_log_pipe[0] = -1;
>> + reopen_log_pipe[1] = -1;
>> +#else
>> if (pipe(reopen_log_pipe)) {
>> barf_perror("pipe");
>> }
>> +#endif
>>
>> /* Setup the database */
>> setup_structure();
>> @@ -1916,6 +1935,7 @@ int main(int argc, char *argv[])
>> fflush(stdout);
>> }
>>
>> +#ifndef __MINIOS__
>> /* redirect to /dev/null now we're ready to accept connections */
>> if (dofork) {
>> int devnull = open("/dev/null", O_RDWR);
>> @@ -1927,6 +1947,7 @@ int main(int argc, char *argv[])
>> close(devnull);
>> xprintf = trace;
>> }
>> +#endif
>
> we could probably handle /dev/null in minios from sys.c
This should really be part of daemonize() except that we still need the
original stdout for outputpid; I'll move it to a finish_daemonize().
>
>> signal(SIGHUP, trigger_reopen_log);
>>
>> @@ -1936,8 +1957,10 @@ int main(int argc, char *argv[])
>> /* Get ready to listen to the tools. */
>> max = initialize_set(&inset, &outset, *sock, *ro_sock, &timeout);
>>
>> +#ifndef __MINIOS__
>> /* Tell the kernel we're up and running. */
>> xenbus_notify_running();
>> +#endif
>>
>> /* Main loop. */
>> for (;;) {
>> @@ -1949,7 +1972,7 @@ int main(int argc, char *argv[])
>> barf_perror("Select failed");
>> }
>>
>> - if (FD_ISSET(reopen_log_pipe[0], &inset)) {
>> + if (reopen_log_pipe[0] != -1 && FD_ISSET(reopen_log_pipe[0], &inset)) {
>> char c;
>> if (read(reopen_log_pipe[0], &c, 1) != 1)
>> barf_perror("read failed");
>> diff --git a/tools/xenstore/xenstored_domain.c b/tools/xenstore/xenstored_domain.c
>> index 661d955..435f76a 100644
>> --- a/tools/xenstore/xenstored_domain.c
>> +++ b/tools/xenstore/xenstored_domain.c
>> @@ -197,10 +197,14 @@ static int destroy_domain(void *_domain)
>> }
>>
>> if (domain->interface) {
>> +#ifdef __MINIOS__
>> + unmap_interface(domain->interface);
>> +#else
>> if (domain->domid == 0)
>> munmap(domain->interface, getpagesize());
>> else
>> unmap_interface(domain->interface);
>> +#endif
>> }
>
> Argh.
> Why dom0 cannot be handled inside unmap_interface?
Because dom0 isn't mapped the same way as domU. The mapping for dom0
is set up by introduce_dom0 which uses mmap directly, so it cannot be
unmapped using the grant table unmap function. When xenstore was using
map_foreign_range this didn't matter, everyone used munmap. In fact,
on Linux it still doesn't matter as the grant table unmap is munmap.
I'll add (as part of #14) a comment here since it's the second time
I've needed to explain this nuance.
>
>> fire_watches(NULL, "@releaseDomain", false);
>> @@ -595,6 +599,12 @@ void restore_existing_connections(void)
>> {
>> }
>>
>> +#ifdef __MINIOS__
>> +static int dom0_init(void)
>> +{
>> + return 0;
>> +}
>> +#else
>> static int dom0_init(void)
>> {
>> evtchn_port_t port;
>> @@ -618,6 +628,7 @@ static int dom0_init(void)
>>
>> return 0;
>> }
>> +#endif
>>
>> void domain_init(void)
>> {
>
^ permalink raw reply [flat|nested] 27+ messages in thread
end of thread, other threads:[~2012-01-25 20:22 UTC | newest]
Thread overview: 27+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-01-25 14:37 [PATCH v4 00/23] Xenstore stub domain Daniel De Graaf
2012-01-25 14:37 ` [PATCH 01/23] xen: reinstate previously unused XENMEM_remove_from_physmap hypercall Daniel De Graaf
2012-01-25 14:37 ` [PATCH 02/23] xen: allow global VIRQ handlers to be delegated to other domains Daniel De Graaf
2012-01-25 14:37 ` [PATCH 03/23] xen: change virq parameters from int to uint32_t Daniel De Graaf
2012-01-25 14:37 ` [PATCH 04/23] xen: use XSM instead of IS_PRIV for getdomaininfo Daniel De Graaf
2012-01-25 14:37 ` [PATCH 05/23] xen: Preserve reserved grant entries when switching versions Daniel De Graaf
2012-01-25 14:37 ` [PATCH 06/23] tools/libxl: pull xenstore/console domids from xenstore Daniel De Graaf
2012-01-25 14:37 ` [PATCH 07/23] lib{xc, xl}: Seed grant tables with xenstore and console grants Daniel De Graaf
2012-01-25 14:37 ` [PATCH 08/23] mini-os: avoid crash if no console is provided Daniel De Graaf
2012-01-25 14:37 ` [PATCH 09/23] mini-os: remove per-fd evtchn limit Daniel De Graaf
2012-01-25 14:37 ` [PATCH 10/23] mini-os: create app-specific configuration Daniel De Graaf
2012-01-25 14:37 ` [PATCH 11/23] mini-os: Move test functions into test.c Daniel De Graaf
2012-01-25 14:37 ` [PATCH 12/23] mini-os: make frontends and xenbus optional Daniel De Graaf
2012-01-25 14:37 ` [PATCH 13/23] mini-os: fix list.h include guard name Daniel De Graaf
2012-01-25 14:37 ` [PATCH 14/23] xenstored: use grant references instead of map_foreign_range Daniel De Graaf
2012-01-25 14:37 ` [PATCH 15/23] xenstored: refactor socket setup code Daniel De Graaf
2012-01-25 14:37 ` [PATCH 16/23] xenstored: add NO_SOCKETS compilation option Daniel De Graaf
2012-01-25 18:52 ` Stefano Stabellini
2012-01-25 14:37 ` [PATCH 17/23] xenstored: support for tdb_copy with TDB_INTERNAL Daniel De Graaf
2012-01-25 14:37 ` [PATCH 18/23] xenstored: support running in minios stubdom Daniel De Graaf
2012-01-25 19:14 ` Stefano Stabellini
2012-01-25 20:22 ` Daniel De Graaf
2012-01-25 14:37 ` [PATCH 19/23] stubdom: enable xenstored build Daniel De Graaf
2012-01-25 14:37 ` [PATCH 20/23] xenstored: add --event parameter for bootstrapping Daniel De Graaf
2012-01-25 14:37 ` [PATCH 21/23] xenstored: use domain_is_unprivileged instead of checking conn->id Daniel De Graaf
2012-01-25 14:37 ` [PATCH 22/23] xenstored: add --priv-domid parameter Daniel De Graaf
2012-01-25 14:37 ` [PATCH 23/23] xenstored: Add stub domain builder Daniel De Graaf
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.