summaryrefslogtreecommitdiff
path: root/tools
diff options
context:
space:
mode:
authorJakub Kicinski <kuba@kernel.org>2022-11-03 11:38:42 -0700
committerJakub Kicinski <kuba@kernel.org>2022-11-03 13:21:54 -0700
commitfbeb229a6622523c092a13c02bd0e15f69240dde (patch)
treea48bf5ff455cedae6555ef071e0d8bdb29487824 /tools
parentd9095f92950bd16745b9ec24ebebc12d14b3a3e8 (diff)
parent9521c9d6a53df9c44a5f5ddbc229ceaf3cf79ef6 (diff)
Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
No conflicts. Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Diffstat (limited to 'tools')
-rw-r--r--tools/iio/iio_utils.c4
-rw-r--r--tools/include/nolibc/string.h17
-rw-r--r--tools/power/pm-graph/README12
-rw-r--r--tools/power/pm-graph/sleepgraph.83
-rwxr-xr-xtools/power/pm-graph/sleepgraph.py225
-rw-r--r--tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c142
6 files changed, 273 insertions, 130 deletions
diff --git a/tools/iio/iio_utils.c b/tools/iio/iio_utils.c
index aadee6d34c74..8d35893b2fa8 100644
--- a/tools/iio/iio_utils.c
+++ b/tools/iio/iio_utils.c
@@ -547,6 +547,10 @@ static int calc_digits(int num)
{
int count = 0;
+ /* It takes a digit to represent zero */
+ if (!num)
+ return 1;
+
while (num != 0) {
num /= 10;
count++;
diff --git a/tools/include/nolibc/string.h b/tools/include/nolibc/string.h
index bef35bee9c44..ad97c0d522b8 100644
--- a/tools/include/nolibc/string.h
+++ b/tools/include/nolibc/string.h
@@ -19,9 +19,9 @@ static __attribute__((unused))
int memcmp(const void *s1, const void *s2, size_t n)
{
size_t ofs = 0;
- char c1 = 0;
+ int c1 = 0;
- while (ofs < n && !(c1 = ((char *)s1)[ofs] - ((char *)s2)[ofs])) {
+ while (ofs < n && !(c1 = ((unsigned char *)s1)[ofs] - ((unsigned char *)s2)[ofs])) {
ofs++;
}
return c1;
@@ -125,14 +125,18 @@ char *strcpy(char *dst, const char *src)
}
/* this function is only used with arguments that are not constants or when
- * it's not known because optimizations are disabled.
+ * it's not known because optimizations are disabled. Note that gcc 12
+ * recognizes an strlen() pattern and replaces it with a jump to strlen(),
+ * thus itself, hence the asm() statement below that's meant to disable this
+ * confusing practice.
*/
static __attribute__((unused))
-size_t nolibc_strlen(const char *str)
+size_t strlen(const char *str)
{
size_t len;
- for (len = 0; str[len]; len++);
+ for (len = 0; str[len]; len++)
+ asm("");
return len;
}
@@ -140,13 +144,12 @@ size_t nolibc_strlen(const char *str)
* the two branches, then will rely on an external definition of strlen().
*/
#if defined(__OPTIMIZE__)
+#define nolibc_strlen(x) strlen(x)
#define strlen(str) ({ \
__builtin_constant_p((str)) ? \
__builtin_strlen((str)) : \
nolibc_strlen((str)); \
})
-#else
-#define strlen(str) nolibc_strlen((str))
#endif
static __attribute__((unused))
diff --git a/tools/power/pm-graph/README b/tools/power/pm-graph/README
index e6020c0d59ec..3213dbe63b74 100644
--- a/tools/power/pm-graph/README
+++ b/tools/power/pm-graph/README
@@ -6,22 +6,22 @@
|_| |___/ |_|
pm-graph: suspend/resume/boot timing analysis tools
- Version: 5.9
+ Version: 5.10
Author: Todd Brandt <todd.e.brandt@intel.com>
- Home Page: https://01.org/pm-graph
+ Home Page: https://www.intel.com/content/www/us/en/developer/topic-technology/open/pm-graph/overview.html
Report bugs/issues at bugzilla.kernel.org Tools/pm-graph
- https://bugzilla.kernel.org/buglist.cgi?component=pm-graph&product=Tools
Full documentation available online & in man pages
- Getting Started:
- https://01.org/pm-graph/documentation/getting-started
+ https://www.intel.com/content/www/us/en/developer/articles/technical/usage.html
- - Config File Format:
- https://01.org/pm-graph/documentation/3-config-file-format
+ - Feature Summary:
+ https://www.intel.com/content/www/us/en/developer/topic-technology/open/pm-graph/features.html
- upstream version in git:
- https://github.com/intel/pm-graph/
+ git clone https://github.com/intel/pm-graph/
Table of Contents
- Overview
diff --git a/tools/power/pm-graph/sleepgraph.8 b/tools/power/pm-graph/sleepgraph.8
index 5126271de98a..643271b6fc6f 100644
--- a/tools/power/pm-graph/sleepgraph.8
+++ b/tools/power/pm-graph/sleepgraph.8
@@ -78,6 +78,9 @@ This helps maintain the consistency of test data for better comparison.
If a wifi connection is available, check that it reconnects after resume. Include
the reconnect time in the total resume time calculation and treat wifi timeouts
as resume failures.
+.TP
+\fB-wifitrace\fR
+Trace through the wifi reconnect time and include it in the timeline.
.SS "advanced"
.TP
diff --git a/tools/power/pm-graph/sleepgraph.py b/tools/power/pm-graph/sleepgraph.py
index 33981adcdd68..cfe343306e08 100755
--- a/tools/power/pm-graph/sleepgraph.py
+++ b/tools/power/pm-graph/sleepgraph.py
@@ -86,7 +86,7 @@ def ascii(text):
# store system values and test parameters
class SystemValues:
title = 'SleepGraph'
- version = '5.9'
+ version = '5.10'
ansi = False
rs = 0
display = ''
@@ -100,6 +100,7 @@ class SystemValues:
ftracelog = False
acpidebug = True
tstat = True
+ wifitrace = False
mindevlen = 0.0001
mincglen = 0.0
cgphase = ''
@@ -124,6 +125,7 @@ class SystemValues:
epath = '/sys/kernel/debug/tracing/events/power/'
pmdpath = '/sys/power/pm_debug_messages'
s0ixpath = '/sys/module/intel_pmc_core/parameters/warn_on_s0ix_failures'
+ s0ixres = '/sys/devices/system/cpu/cpuidle/low_power_idle_system_residency_us'
acpipath='/sys/module/acpi/parameters/debug_level'
traceevents = [
'suspend_resume',
@@ -180,6 +182,7 @@ class SystemValues:
tmstart = 'SUSPEND START %Y%m%d-%H:%M:%S.%f'
tmend = 'RESUME COMPLETE %Y%m%d-%H:%M:%S.%f'
tracefuncs = {
+ 'async_synchronize_full': {},
'sys_sync': {},
'ksys_sync': {},
'__pm_notifier_call_chain': {},
@@ -304,6 +307,7 @@ class SystemValues:
[2, 'suspendstats', 'sh', '-c', 'grep -v invalid /sys/power/suspend_stats/*'],
[2, 'cpuidle', 'sh', '-c', 'grep -v invalid /sys/devices/system/cpu/cpu*/cpuidle/state*/s2idle/*'],
[2, 'battery', 'sh', '-c', 'grep -v invalid /sys/class/power_supply/*/*'],
+ [2, 'thermal', 'sh', '-c', 'grep . /sys/class/thermal/thermal_zone*/temp'],
]
cgblacklist = []
kprobes = dict()
@@ -777,7 +781,7 @@ class SystemValues:
return
if not quiet:
sysvals.printSystemInfo(False)
- pprint('INITIALIZING FTRACE...')
+ pprint('INITIALIZING FTRACE')
# turn trace off
self.fsetVal('0', 'tracing_on')
self.cleanupFtrace()
@@ -841,7 +845,7 @@ class SystemValues:
for name in self.dev_tracefuncs:
self.defaultKprobe(name, self.dev_tracefuncs[name])
if not quiet:
- pprint('INITIALIZING KPROBES...')
+ pprint('INITIALIZING KPROBES')
self.addKprobes(self.verbose)
if(self.usetraceevents):
# turn trace events on
@@ -1133,6 +1137,15 @@ class SystemValues:
self.cfgdef[file] = fp.read().strip()
fp.write(value)
fp.close()
+ def s0ixSupport(self):
+ if not os.path.exists(self.s0ixres) or not os.path.exists(self.mempowerfile):
+ return False
+ fp = open(sysvals.mempowerfile, 'r')
+ data = fp.read().strip()
+ fp.close()
+ if '[s2idle]' in data:
+ return True
+ return False
def haveTurbostat(self):
if not self.tstat:
return False
@@ -1146,7 +1159,7 @@ class SystemValues:
self.vprint(out)
return True
return False
- def turbostat(self):
+ def turbostat(self, s0ixready):
cmd = self.getExec('turbostat')
rawout = keyline = valline = ''
fullcmd = '%s -q -S echo freeze > %s' % (cmd, self.powerfile)
@@ -1173,6 +1186,8 @@ class SystemValues:
for key in keyline:
idx = keyline.index(key)
val = valline[idx]
+ if key == 'SYS%LPI' and not s0ixready and re.match('^[0\.]*$', val):
+ continue
out.append('%s=%s' % (key, val))
return '|'.join(out)
def netfixon(self, net='both'):
@@ -1183,14 +1198,6 @@ class SystemValues:
out = ascii(fp.read()).strip()
fp.close()
return out
- def wifiRepair(self):
- out = self.netfixon('wifi')
- if not out or 'error' in out.lower():
- return ''
- m = re.match('WIFI \S* ONLINE (?P<action>\S*)', out)
- if not m:
- return 'dead'
- return m.group('action')
def wifiDetails(self, dev):
try:
info = open('/sys/class/net/%s/device/uevent' % dev, 'r').read().strip()
@@ -1220,11 +1227,6 @@ class SystemValues:
return '%s reconnected %.2f' % \
(self.wifiDetails(dev), max(0, time.time() - start))
time.sleep(0.01)
- if self.netfix:
- res = self.wifiRepair()
- if res:
- timeout = max(0, time.time() - start)
- return '%s %s %d' % (self.wifiDetails(dev), res, timeout)
return '%s timeout %d' % (self.wifiDetails(dev), timeout)
def errorSummary(self, errinfo, msg):
found = False
@@ -1346,6 +1348,20 @@ class SystemValues:
for i in self.rslist:
self.setVal(self.rstgt, i)
pprint('runtime suspend settings restored on %d devices' % len(self.rslist))
+ def start(self, pm):
+ if self.useftrace:
+ self.dlog('start ftrace tracing')
+ self.fsetVal('1', 'tracing_on')
+ if self.useprocmon:
+ self.dlog('start the process monitor')
+ pm.start()
+ def stop(self, pm):
+ if self.useftrace:
+ if self.useprocmon:
+ self.dlog('stop the process monitor')
+ pm.stop()
+ self.dlog('stop ftrace tracing')
+ self.fsetVal('0', 'tracing_on')
sysvals = SystemValues()
switchvalues = ['enable', 'disable', 'on', 'off', 'true', 'false', '1', '0']
@@ -1643,19 +1659,20 @@ class Data:
ubiquitous = False
if kprobename in dtf and 'ub' in dtf[kprobename]:
ubiquitous = True
- title = cdata+' '+rdata
- mstr = '\(.*\) *(?P<args>.*) *\((?P<caller>.*)\+.* arg1=(?P<ret>.*)'
- m = re.match(mstr, title)
- if m:
- c = m.group('caller')
- a = m.group('args').strip()
- r = m.group('ret')
+ mc = re.match('\(.*\) *(?P<args>.*)', cdata)
+ mr = re.match('\((?P<caller>\S*).* arg1=(?P<ret>.*)', rdata)
+ if mc and mr:
+ c = mr.group('caller').split('+')[0]
+ a = mc.group('args').strip()
+ r = mr.group('ret')
if len(r) > 6:
r = ''
else:
r = 'ret=%s ' % r
if ubiquitous and c in dtf and 'ub' in dtf[c]:
return False
+ else:
+ return False
color = sysvals.kprobeColor(kprobename)
e = DevFunction(displayname, a, c, r, start, end, ubiquitous, proc, pid, color)
tgtdev['src'].append(e)
@@ -1772,6 +1789,14 @@ class Data:
e.time = self.trimTimeVal(e.time, t0, dT, left)
e.end = self.trimTimeVal(e.end, t0, dT, left)
e.length = e.end - e.time
+ if('cpuexec' in d):
+ cpuexec = dict()
+ for e in d['cpuexec']:
+ c0, cN = e
+ c0 = self.trimTimeVal(c0, t0, dT, left)
+ cN = self.trimTimeVal(cN, t0, dT, left)
+ cpuexec[(c0, cN)] = d['cpuexec'][e]
+ d['cpuexec'] = cpuexec
for dir in ['suspend', 'resume']:
list = []
for e in self.errorinfo[dir]:
@@ -2086,75 +2111,43 @@ class Data:
return d
def addProcessUsageEvent(self, name, times):
# get the start and end times for this process
- maxC = 0
- tlast = 0
- start = -1
- end = -1
+ cpuexec = dict()
+ tlast = start = end = -1
for t in sorted(times):
- if tlast == 0:
+ if tlast < 0:
tlast = t
continue
- if name in self.pstl[t]:
- if start == -1 or tlast < start:
+ if name in self.pstl[t] and self.pstl[t][name] > 0:
+ if start < 0:
start = tlast
- if end == -1 or t > end:
- end = t
+ end, key = t, (tlast, t)
+ maxj = (t - tlast) * 1024.0
+ cpuexec[key] = min(1.0, float(self.pstl[t][name]) / maxj)
tlast = t
- if start == -1 or end == -1:
- return 0
+ if start < 0 or end < 0:
+ return
# add a new action for this process and get the object
out = self.newActionGlobal(name, start, end, -3)
- if not out:
- return 0
- phase, devname = out
- dev = self.dmesg[phase]['list'][devname]
- # get the cpu exec data
- tlast = 0
- clast = 0
- cpuexec = dict()
- for t in sorted(times):
- if tlast == 0 or t <= start or t > end:
- tlast = t
- continue
- list = self.pstl[t]
- c = 0
- if name in list:
- c = list[name]
- if c > maxC:
- maxC = c
- if c != clast:
- key = (tlast, t)
- cpuexec[key] = c
- tlast = t
- clast = c
- dev['cpuexec'] = cpuexec
- return maxC
+ if out:
+ phase, devname = out
+ dev = self.dmesg[phase]['list'][devname]
+ dev['cpuexec'] = cpuexec
def createProcessUsageEvents(self):
- # get an array of process names
- proclist = []
- for t in sorted(self.pstl):
- pslist = self.pstl[t]
- for ps in sorted(pslist):
- if ps not in proclist:
- proclist.append(ps)
- # get a list of data points for suspend and resume
- tsus = []
- tres = []
+ # get an array of process names and times
+ proclist = {'sus': dict(), 'res': dict()}
+ tdata = {'sus': [], 'res': []}
for t in sorted(self.pstl):
- if t < self.tSuspended:
- tsus.append(t)
- else:
- tres.append(t)
+ dir = 'sus' if t < self.tSuspended else 'res'
+ for ps in sorted(self.pstl[t]):
+ if ps not in proclist[dir]:
+ proclist[dir][ps] = 0
+ tdata[dir].append(t)
# process the events for suspend and resume
- if len(proclist) > 0:
+ if len(proclist['sus']) > 0 or len(proclist['res']) > 0:
sysvals.vprint('Process Execution:')
- for ps in proclist:
- c = self.addProcessUsageEvent(ps, tsus)
- if c > 0:
- sysvals.vprint('%25s (sus): %d' % (ps, c))
- c = self.addProcessUsageEvent(ps, tres)
- if c > 0:
- sysvals.vprint('%25s (res): %d' % (ps, c))
+ for dir in ['sus', 'res']:
+ for ps in sorted(proclist[dir]):
+ self.addProcessUsageEvent(ps, tdata[dir])
def handleEndMarker(self, time, msg=''):
dm = self.dmesg
self.setEnd(time, msg)
@@ -3218,7 +3211,7 @@ class ProcessMonitor:
# markers, and/or kprobes required for primary parsing.
def doesTraceLogHaveTraceEvents():
kpcheck = ['_cal: (', '_ret: (']
- techeck = ['suspend_resume', 'device_pm_callback']
+ techeck = ['suspend_resume', 'device_pm_callback', 'tracing_mark_write']
tmcheck = ['SUSPEND START', 'RESUME COMPLETE']
sysvals.usekprobes = False
fp = sysvals.openlog(sysvals.ftracefile, 'r')
@@ -3241,7 +3234,7 @@ def doesTraceLogHaveTraceEvents():
check.remove(i)
tmcheck = check
fp.close()
- sysvals.usetraceevents = True if len(techeck) < 2 else False
+ sysvals.usetraceevents = True if len(techeck) < 3 else False
sysvals.usetracemarkers = True if len(tmcheck) == 0 else False
# Function: appendIncompleteTraceLog
@@ -3456,6 +3449,8 @@ def parseTraceLog(live=False):
continue
# process cpu exec line
if t.type == 'tracing_mark_write':
+ if t.name == 'CMD COMPLETE' and data.tKernRes == 0:
+ data.tKernRes = t.time
m = re.match(tp.procexecfmt, t.name)
if(m):
parts, msg = 1, m.group('ps')
@@ -3674,6 +3669,9 @@ def parseTraceLog(live=False):
e = next((x for x in reversed(tp.ktemp[key]) if x['end'] < 0), 0)
if not e:
continue
+ if (t.time - e['begin']) * 1000 < sysvals.mindevlen:
+ tp.ktemp[key].pop()
+ continue
e['end'] = t.time
e['rdata'] = kprobedata
# end of kernel resume
@@ -4213,6 +4211,8 @@ def callgraphHTML(sv, hf, num, cg, title, color, devid):
fmt = '<n>(%.3f ms @ '+sv.timeformat+')</n>'
flen = fmt % (line.length*1000, line.time)
if line.isLeaf():
+ if line.length * 1000 < sv.mincglen:
+ continue
hf.write(html_func_leaf.format(line.name, flen))
elif line.freturn:
hf.write(html_func_end)
@@ -4827,14 +4827,11 @@ def createHTML(testruns, testfail):
if('cpuexec' in dev):
for t in sorted(dev['cpuexec']):
start, end = t
- j = float(dev['cpuexec'][t]) / 5
- if j > 1.0:
- j = 1.0
height = '%.3f' % (rowheight/3)
top = '%.3f' % (rowtop + devtl.scaleH + 2*rowheight/3)
left = '%f' % (((start-m0)*100)/mTotal)
width = '%f' % ((end-start)*100/mTotal)
- color = 'rgba(255, 0, 0, %f)' % j
+ color = 'rgba(255, 0, 0, %f)' % dev['cpuexec'][t]
devtl.html += \
html_cpuexec.format(left, top, height, width, color)
if('src' not in dev):
@@ -5453,17 +5450,9 @@ def executeSuspend(quiet=False):
call('sync', shell=True)
sv.dlog('read dmesg')
sv.initdmesg()
- # start ftrace
- if sv.useftrace:
- if not quiet:
- pprint('START TRACING')
- sv.dlog('start ftrace tracing')
- sv.fsetVal('1', 'tracing_on')
- if sv.useprocmon:
- sv.dlog('start the process monitor')
- pm.start()
- sv.dlog('run the cmdinfo list before')
+ sv.dlog('cmdinfo before')
sv.cmdinfo(True)
+ sv.start(pm)
# execute however many s/r runs requested
for count in range(1,sv.execcount+1):
# x2delay in between test runs
@@ -5500,6 +5489,7 @@ def executeSuspend(quiet=False):
if res != 0:
tdata['error'] = 'cmd returned %d' % res
else:
+ s0ixready = sv.s0ixSupport()
mode = sv.suspendmode
if sv.memmode and os.path.exists(sv.mempowerfile):
mode = 'mem'
@@ -5509,9 +5499,10 @@ def executeSuspend(quiet=False):
sv.testVal(sv.diskpowerfile, 'radio', sv.diskmode)
if sv.acpidebug:
sv.testVal(sv.acpipath, 'acpi', '0xe')
- if mode == 'freeze' and sv.haveTurbostat():
+ if ((mode == 'freeze') or (sv.memmode == 's2idle')) \
+ and sv.haveTurbostat():
# execution will pause here
- turbo = sv.turbostat()
+ turbo = sv.turbostat(s0ixready)
if turbo:
tdata['turbo'] = turbo
else:
@@ -5522,7 +5513,8 @@ def executeSuspend(quiet=False):
pf.close()
except Exception as e:
tdata['error'] = str(e)
- sv.dlog('system returned from resume')
+ sv.fsetVal('CMD COMPLETE', 'trace_marker')
+ sv.dlog('system returned')
# reset everything
sv.testVal('restoreall')
if(sv.rtcwake):
@@ -5535,33 +5527,29 @@ def executeSuspend(quiet=False):
sv.fsetVal('WAIT END', 'trace_marker')
# return from suspend
pprint('RESUME COMPLETE')
- sv.fsetVal(datetime.now().strftime(sv.tmend), 'trace_marker')
+ if(count < sv.execcount):
+ sv.fsetVal(datetime.now().strftime(sv.tmend), 'trace_marker')
+ elif(not sv.wifitrace):
+ sv.fsetVal(datetime.now().strftime(sv.tmend), 'trace_marker')
+ sv.stop(pm)
if sv.wifi and wifi:
tdata['wifi'] = sv.pollWifi(wifi)
sv.dlog('wifi check, %s' % tdata['wifi'])
- if sv.netfix:
- netfixout = sv.netfixon('wired')
- elif sv.netfix:
- netfixout = sv.netfixon()
- if sv.netfix and netfixout:
- tdata['netfix'] = netfixout
+ if(count == sv.execcount and sv.wifitrace):
+ sv.fsetVal(datetime.now().strftime(sv.tmend), 'trace_marker')
+ sv.stop(pm)
+ if sv.netfix:
+ tdata['netfix'] = sv.netfixon()
sv.dlog('netfix, %s' % tdata['netfix'])
if(sv.suspendmode == 'mem' or sv.suspendmode == 'command'):
sv.dlog('read the ACPI FPDT')
tdata['fw'] = getFPDT(False)
testdata.append(tdata)
- sv.dlog('run the cmdinfo list after')
+ sv.dlog('cmdinfo after')
cmdafter = sv.cmdinfo(False)
- # stop ftrace
- if sv.useftrace:
- if sv.useprocmon:
- sv.dlog('stop the process monitor')
- pm.stop()
- sv.fsetVal('0', 'tracing_on')
# grab a copy of the dmesg output
if not quiet:
pprint('CAPTURING DMESG')
- sysvals.dlog('EXECUTION TRACE END')
sv.getdmesg(testdata)
# grab a copy of the ftrace output
if sv.useftrace:
@@ -6350,6 +6338,8 @@ def data_from_html(file, outpath, issues, fulldetail=False):
if not m:
continue
name, time, phase = m.group('n'), m.group('t'), m.group('p')
+ if name == 'async_synchronize_full':
+ continue
if ' async' in name or ' sync' in name:
name = ' '.join(name.split(' ')[:-1])
if phase.startswith('suspend'):
@@ -6701,6 +6691,7 @@ def printHelp():
' -skiphtml Run the test and capture the trace logs, but skip the timeline (default: disabled)\n'\
' -result fn Export a results table to a text file for parsing.\n'\
' -wifi If a wifi connection is available, check that it reconnects after resume.\n'\
+ ' -wifitrace Trace kernel execution through wifi reconnect.\n'\
' -netfix Use netfix to reset the network in the event it fails to resume.\n'\
' [testprep]\n'\
' -sync Sync the filesystems before starting the test\n'\
@@ -6828,6 +6819,8 @@ if __name__ == '__main__':
sysvals.sync = True
elif(arg == '-wifi'):
sysvals.wifi = True
+ elif(arg == '-wifitrace'):
+ sysvals.wifitrace = True
elif(arg == '-netfix'):
sysvals.netfix = True
elif(arg == '-gzip'):
diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
index 8a5cb800f50e..2a5727188c8d 100644
--- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
@@ -15,9 +15,13 @@
#include <time.h>
#include <sched.h>
#include <signal.h>
+#include <pthread.h>
#include <sys/eventfd.h>
+/* Defined in include/linux/kvm_types.h */
+#define GPA_INVALID (~(ulong)0)
+
#define SHINFO_REGION_GVA 0xc0000000ULL
#define SHINFO_REGION_GPA 0xc0000000ULL
#define SHINFO_REGION_SLOT 10
@@ -44,6 +48,8 @@
#define MIN_STEAL_TIME 50000
+#define SHINFO_RACE_TIMEOUT 2 /* seconds */
+
#define __HYPERVISOR_set_timer_op 15
#define __HYPERVISOR_sched_op 29
#define __HYPERVISOR_event_channel_op 32
@@ -126,7 +132,7 @@ struct {
struct kvm_irq_routing_entry entries[2];
} irq_routes;
-bool guest_saw_irq;
+static volatile bool guest_saw_irq;
static void evtchn_handler(struct ex_regs *regs)
{
@@ -148,6 +154,7 @@ static void guest_wait_for_irq(void)
static void guest_code(void)
{
struct vcpu_runstate_info *rs = (void *)RUNSTATE_VADDR;
+ int i;
__asm__ __volatile__(
"sti\n"
@@ -325,6 +332,49 @@ static void guest_code(void)
guest_wait_for_irq();
GUEST_SYNC(21);
+ /* Racing host ioctls */
+
+ guest_wait_for_irq();
+
+ GUEST_SYNC(22);
+ /* Racing vmcall against host ioctl */
+
+ ports[0] = 0;
+
+ p = (struct sched_poll) {
+ .ports = ports,
+ .nr_ports = 1,
+ .timeout = 0
+ };
+
+wait_for_timer:
+ /*
+ * Poll for a timer wake event while the worker thread is mucking with
+ * the shared info. KVM XEN drops timer IRQs if the shared info is
+ * invalid when the timer expires. Arbitrarily poll 100 times before
+ * giving up and asking the VMM to re-arm the timer. 100 polls should
+ * consume enough time to beat on KVM without taking too long if the
+ * timer IRQ is dropped due to an invalid event channel.
+ */
+ for (i = 0; i < 100 && !guest_saw_irq; i++)
+ asm volatile("vmcall"
+ : "=a" (rax)
+ : "a" (__HYPERVISOR_sched_op),
+ "D" (SCHEDOP_poll),
+ "S" (&p)
+ : "memory");
+
+ /*
+ * Re-send the timer IRQ if it was (likely) dropped due to the timer
+ * expiring while the event channel was invalid.
+ */
+ if (!guest_saw_irq) {
+ GUEST_SYNC(23);
+ goto wait_for_timer;
+ }
+ guest_saw_irq = false;
+
+ GUEST_SYNC(24);
}
static int cmp_timespec(struct timespec *a, struct timespec *b)
@@ -352,11 +402,36 @@ static void handle_alrm(int sig)
TEST_FAIL("IRQ delivery timed out");
}
+static void *juggle_shinfo_state(void *arg)
+{
+ struct kvm_vm *vm = (struct kvm_vm *)arg;
+
+ struct kvm_xen_hvm_attr cache_init = {
+ .type = KVM_XEN_ATTR_TYPE_SHARED_INFO,
+ .u.shared_info.gfn = SHINFO_REGION_GPA / PAGE_SIZE
+ };
+
+ struct kvm_xen_hvm_attr cache_destroy = {
+ .type = KVM_XEN_ATTR_TYPE_SHARED_INFO,
+ .u.shared_info.gfn = GPA_INVALID
+ };
+
+ for (;;) {
+ __vm_ioctl(vm, KVM_XEN_HVM_SET_ATTR, &cache_init);
+ __vm_ioctl(vm, KVM_XEN_HVM_SET_ATTR, &cache_destroy);
+ pthread_testcancel();
+ };
+
+ return NULL;
+}
+
int main(int argc, char *argv[])
{
struct timespec min_ts, max_ts, vm_ts;
struct kvm_vm *vm;
+ pthread_t thread;
bool verbose;
+ int ret;
verbose = argc > 1 && (!strncmp(argv[1], "-v", 3) ||
!strncmp(argv[1], "--verbose", 10));
@@ -785,6 +860,71 @@ int main(int argc, char *argv[])
case 21:
TEST_ASSERT(!evtchn_irq_expected,
"Expected event channel IRQ but it didn't happen");
+ alarm(0);
+
+ if (verbose)
+ printf("Testing shinfo lock corruption (KVM_XEN_HVM_EVTCHN_SEND)\n");
+
+ ret = pthread_create(&thread, NULL, &juggle_shinfo_state, (void *)vm);
+ TEST_ASSERT(ret == 0, "pthread_create() failed: %s", strerror(ret));
+
+ struct kvm_irq_routing_xen_evtchn uxe = {
+ .port = 1,
+ .vcpu = vcpu->id,
+ .priority = KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL
+ };
+
+ evtchn_irq_expected = true;
+ for (time_t t = time(NULL) + SHINFO_RACE_TIMEOUT; time(NULL) < t;)
+ __vm_ioctl(vm, KVM_XEN_HVM_EVTCHN_SEND, &uxe);
+ break;
+
+ case 22:
+ TEST_ASSERT(!evtchn_irq_expected,
+ "Expected event channel IRQ but it didn't happen");
+
+ if (verbose)
+ printf("Testing shinfo lock corruption (SCHEDOP_poll)\n");
+
+ shinfo->evtchn_pending[0] = 1;
+
+ evtchn_irq_expected = true;
+ tmr.u.timer.expires_ns = rs->state_entry_time +
+ SHINFO_RACE_TIMEOUT * 1000000000ULL;
+ vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &tmr);
+ break;
+
+ case 23:
+ /*
+ * Optional and possibly repeated sync point.
+ * Injecting the timer IRQ may fail if the
+ * shinfo is invalid when the timer expires.
+ * If the timer has expired but the IRQ hasn't
+ * been delivered, rearm the timer and retry.
+ */
+ vcpu_ioctl(vcpu, KVM_XEN_VCPU_GET_ATTR, &tmr);
+
+ /* Resume the guest if the timer is still pending. */
+ if (tmr.u.timer.expires_ns)
+ break;
+
+ /* All done if the IRQ was delivered. */
+ if (!evtchn_irq_expected)
+ break;
+
+ tmr.u.timer.expires_ns = rs->state_entry_time +
+ SHINFO_RACE_TIMEOUT * 1000000000ULL;
+ vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &tmr);
+ break;
+ case 24:
+ TEST_ASSERT(!evtchn_irq_expected,
+ "Expected event channel IRQ but it didn't happen");
+
+ ret = pthread_cancel(thread);
+ TEST_ASSERT(ret == 0, "pthread_cancel() failed: %s", strerror(ret));
+
+ ret = pthread_join(thread, 0);
+ TEST_ASSERT(ret == 0, "pthread_join() failed: %s", strerror(ret));
goto done;
case 0x20: