From: Ian Jackson <ian.jackson@eu.citrix.com>
To: xen-devel@lists.xensource.com
Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>,
Wei Liu <wei.liu2@citrix.com>,
Ian Campbell <Ian.Campbell@citrix.com>,
Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Subject: [PATCH 14/28] libxl: emuids: Pass emuid to dm destruction
Date: Tue, 22 Dec 2015 18:44:49 +0000 [thread overview]
Message-ID: <1450809903-3393-15-git-send-email-ian.jackson@eu.citrix.com> (raw)
In-Reply-To: <1450809903-3393-1-git-send-email-ian.jackson@eu.citrix.com>
Rather than recomputing this with a switch() statement, etc., we rely
on the emuidmap to tell us which qemus to destroy.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
v6: New patch. This is a new approach in v6.
squash! libxl: Pass emuid to dm destruction
---
tools/libxl/libxl.c | 31 ++++++++++++-------------------
tools/libxl/libxl_dm.c | 13 +++++++++----
tools/libxl/libxl_internal.h | 3 ++-
3 files changed, 23 insertions(+), 24 deletions(-)
diff --git a/tools/libxl/libxl.c b/tools/libxl/libxl.c
index 2a0c092..dff4478 100644
--- a/tools/libxl/libxl.c
+++ b/tools/libxl/libxl.c
@@ -1606,7 +1606,8 @@ void libxl__destroy_domid(libxl__egc *egc, libxl__destroy_domid_state *dis)
libxl_ctx *ctx = CTX;
uint32_t domid = dis->domid;
char *dom_path;
- int rc, dm_present;
+ int rc;
+ unsigned emuidmap;
libxl__ev_child_init(&dis->destroyer);
@@ -1633,27 +1634,19 @@ void libxl__destroy_domid(libxl__egc *egc, libxl__destroy_domid_state *dis)
goto out;
}
- switch (libxl__domain_type(gc, domid)) {
- case LIBXL_DOMAIN_TYPE_HVM:
- if (libxl_get_stubdom_id(CTX, domid)) {
- dm_present = 0;
- break;
- }
- /* fall through */
- case LIBXL_DOMAIN_TYPE_PV:
- dm_present = 1;
- break;
- case LIBXL_DOMAIN_TYPE_INVALID:
- rc = ERROR_FAIL;
- goto out;
- default:
- abort();
- }
+ libxl__dm_emuidmap_get_bodgeerrors(gc, domid, &emuidmap);
- if (dm_present) {
- libxl__destroy_device_model(gc, domid);
+ libxl__destroy_device_model(gc, domid, emuidmap, EMUID_PV);
+ if (libxl__domain_type(gc, domid) == LIBXL_DOMAIN_TYPE_HVM &&
+ libxl_get_stubdom_id(CTX, domid)) {
+ /* This is destroyed in libxl__domain_destroy */
+ } else {
+ libxl__destroy_device_model(gc, domid, emuidmap, EMUID_DM);
}
+
+ libxl__destroy_device_model(gc, domid, emuidmap, EMUID_PV);
+
dis->drs.ao = ao;
dis->drs.domid = domid;
dis->drs.callback = devices_destroy_cb;
diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
index 2bd1eb0..bcae664 100644
--- a/tools/libxl/libxl_dm.c
+++ b/tools/libxl/libxl_dm.c
@@ -2111,20 +2111,25 @@ out:
return rc;
}
-int libxl__destroy_device_model(libxl__gc *gc, uint32_t domid)
+int libxl__destroy_device_model(libxl__gc *gc, uint32_t domid,
+ unsigned emuidmap, int emuid)
{
int rc;
+ if (!(emuidmap & (1u << emuid)))
+ return 0;
+
char *path = libxl__device_model_xs_path(gc, LIBXL_TOOLSTACK_DOMID,
- domid, EMUID_DM, "");
+ domid, emuid, "");
if (!xs_rm(CTX->xsh, XBT_NULL, path))
LOG(ERROR, "xs_rm failed for %s", path);
+
/* We should try to destroy the device model anyway. */
rc = kill_device_model(gc,
GCSPRINTF("/local/domain/%d/image/device-model-pid", domid));
if (rc)
- LOG(ERROR, "libxl__destroy_device_model failed for %d",
- domid);
+ LOG(ERROR, "libxl__destroy_device_model failed for %d (emuid=%d)",
+ domid, emuid);
libxl__qmp_cleanup(gc, domid);
return rc;
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
index 0194823..4a9003d 100644
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -1656,7 +1656,8 @@ _hidden int libxl__wait_for_device_model_deprecated(libxl__gc *gc,
void *userdata),
void *check_callback_userdata);
-_hidden int libxl__destroy_device_model(libxl__gc *gc, uint32_t domid);
+_hidden int libxl__destroy_device_model(libxl__gc *gc, uint32_t domid,
+ unsigned emuidmap, int emuid);
_hidden const libxl_vnc_info *libxl__dm_vnc(const libxl_domain_config *g_cfg);
--
1.7.10.4
next prev parent reply other threads:[~2015-12-22 18:44 UTC|newest]
Thread overview: 61+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-22 18:44 [RFC PATCH v6 00/28] libxl: Deprivilege qemu Ian Jackson
2015-12-22 18:44 ` [PATCH 01/28] libxl: Move FILLZERO up in libxl_internal.h Ian Jackson
2016-01-07 17:08 ` Ian Campbell
2015-12-22 18:44 ` [PATCH 02/28] libxl: libxl_types_internal.idl: Add Emacs mode comment Ian Jackson
2016-01-07 17:09 ` Ian Campbell
2015-12-22 18:44 ` [PATCH 03/28] libxl: Provide libxl__dm_support_* Ian Jackson
2016-01-07 17:13 ` Ian Campbell
2016-01-08 12:13 ` Ian Jackson
2016-01-11 17:00 ` Jim Fehlig
2016-01-14 10:14 ` Ian Campbell
2016-01-14 18:31 ` Jim Fehlig
2016-01-15 9:56 ` Ian Campbell
2016-01-15 14:54 ` Jim Fehlig
2015-12-22 18:44 ` [PATCH 04/28] libxl: Invoke libxl__dm_support_* Ian Jackson
2015-12-22 18:44 ` [PATCH 05/28] libxl: libxl__spawn_stub_dm: Introduce `dmpath' Ian Jackson
2015-12-22 18:44 ` [PATCH 06/28] libxl: qemu_pci_*: Introduce DMPATH local macro, twice Ian Jackson
2015-12-22 18:44 ` [PATCH 07/28] libxl: libxl__device_model_xs_path: Add emulator_id parameter Ian Jackson
2015-12-22 18:44 ` [PATCH 08/28] libxl: libxl__destroy_domid: Bring dm destruction together Ian Jackson
2015-12-22 18:44 ` [PATCH 09/28] libxl: Move some error handling and cleanup into libxl__destroy_device_model Ian Jackson
2015-12-22 18:44 ` [PATCH 10/28] libxl: kill_device_model: Silently tolerate ENOENT on pid xs path Ian Jackson
2015-12-22 18:44 ` [PATCH 11/28] libxl: emuids: Pass correct emuid to internal functions Ian Jackson
2015-12-22 18:44 ` [PATCH 12/28] libxl: Use libxl__device_model_xs_path in libxl__spawn_qdisk_backend Ian Jackson
2015-12-22 18:44 ` [PATCH 13/28] libxl: emuids: Record which emuids we have started to create Ian Jackson
2015-12-22 18:44 ` Ian Jackson [this message]
2015-12-22 18:44 ` [PATCH 15/28] libxl: emuids: Pass emuid to device model argument construction Ian Jackson
2015-12-22 18:44 ` [PATCH 16/28] libxl: emuids: Provide libxl__dm_xs_path_rel Ian Jackson
2015-12-22 18:44 ` [PATCH 17/28] libxl: emuids: Do not open-code device-model/%u in libxl__destroy_qdisk_backend Ian Jackson
2015-12-22 18:44 ` [PATCH 18/28] libxl: emuids: Change pid path in xenstore Ian Jackson
2015-12-22 18:44 ` [PATCH 19/28] libxl: Improve libxl__destroy_device_model Ian Jackson
2015-12-22 18:44 ` [PATCH 20/28] libxl: domcreate_dm_support_checked: Introduce `goto out' Ian Jackson
2015-12-22 18:44 ` [PATCH 21/28] libxl: dm user: Reject attempts to set user!=root with qemu trad Ian Jackson
2016-01-07 17:20 ` Ian Campbell
2016-01-08 12:16 ` Ian Jackson
2016-01-08 12:23 ` Ian Campbell
2015-12-22 18:44 ` [PATCH 22/28] libxl: dm user: Document the default Ian Jackson
2016-01-07 17:20 ` Ian Campbell
2015-12-22 18:44 ` [PATCH 23/28] libxl: dm user: Move user choice earlier, and fill in config Ian Jackson
2015-12-22 18:44 ` [PATCH 24/28] libxl: dm spawn records rc in state struct rather than passing as argument Ian Jackson
2015-12-22 18:45 ` [PATCH 25/28] libxl: emuids: Perhaps change dm xs control path Ian Jackson
2016-01-07 17:26 ` Ian Campbell
2016-01-08 14:12 ` Ian Jackson
2016-01-08 14:36 ` Ian Campbell
2016-01-08 14:45 ` Ian Jackson
2016-01-08 14:49 ` Ian Campbell
2015-12-22 18:45 ` [PATCH 26/28] libxl: spawns two QEMUs for HVM guests Ian Jackson
2016-01-07 17:28 ` Ian Campbell
2016-01-08 14:35 ` Ian Jackson
2016-01-08 14:52 ` Ian Campbell
2015-12-22 18:45 ` [PATCH 27/28] libxl: Limit qemu physmap entries Ian Jackson
2016-01-07 17:28 ` Ian Campbell
2015-12-22 18:45 ` [PATCH 28/28] libxl: xsrestrict QEMU Ian Jackson
2016-01-07 17:36 ` Ian Campbell
2016-01-08 14:38 ` Ian Jackson
2016-04-10 19:52 ` Stefano Stabellini
2016-01-07 16:19 ` [RFC PATCH v6 00/28] libxl: Deprivilege qemu Wei Liu
2016-01-07 16:23 ` Stefano Stabellini
2016-01-07 16:36 ` Ian Jackson
2016-04-10 19:36 ` Stefano Stabellini
2016-04-11 10:35 ` Wei Liu
2016-04-14 17:27 ` Ian Jackson
2016-04-28 14:32 ` Ian Jackson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1450809903-3393-15-git-send-email-ian.jackson@eu.citrix.com \
--to=ian.jackson@eu.citrix.com \
--cc=Ian.Campbell@citrix.com \
--cc=stefano.stabellini@eu.citrix.com \
--cc=wei.liu2@citrix.com \
--cc=xen-devel@lists.xensource.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).