[PATCH 0/3] tests: three easy Coverity fixes

[PATCH 0/3] tests: three easy Coverity fixes

[Peter Maydell]

We recently started running Coverity on our tests/ code,
which has revealed a pile of issues that we were previously
blissfully unaware of. This series fixes three easy ones:
 * use of free() where g_free() was meant
 * unnecessary NULL check
 * side effect inside g_assert() argument

thanks
-- PMM

Peter Maydell (3):
  tests/qtest/tpm-util.c: Free memory with correct free function
  tests/qtest/rtc-test: Remove pointless NULL check
  tests: Avoid side effects inside g_assert() arguments

 tests/qtest/ipmi-bt-test.c  | 6 ++++--
 tests/qtest/ipmi-kcs-test.c | 3 ++-
 tests/qtest/rtc-test.c      | 6 ++----
 tests/qtest/tpm-util.c      | 4 ++--
 4 files changed, 10 insertions(+), 9 deletions(-)

-- 
2.20.1

[PATCH 1/3] tests/qtest/tpm-util.c: Free memory with correct free function

[Peter Maydell]

tpm_util_migration_start_qemu() allocates memory with g_strdup_printf()
but frees it with free() rather than g_free(), which provokes Coverity
complaints (CID 1432379, 1432350). Use the correct free function.

Fixes: Coverity CID 1432379, CID 1432350
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
With newer glib (2.46 and up) g_free() is guaranteed to be the same
as free(), but matching things up is neater anyway.
---
 tests/qtest/tpm-util.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/qtest/tpm-util.c b/tests/qtest/tpm-util.c
index b70cc32d600..3a40ff3f96c 100644
--- a/tests/qtest/tpm-util.c
+++ b/tests/qtest/tpm-util.c
@@ -289,6 +289,6 @@ void tpm_util_migration_start_qemu(QTestState **src_qemu,
 
     *dst_qemu = qtest_init(dst_qemu_args);
 
-    free(src_qemu_args);
-    free(dst_qemu_args);
+    g_free(src_qemu_args);
+    g_free(dst_qemu_args);
 }
-- 
2.20.1

Re: [PATCH 1/3] tests/qtest/tpm-util.c: Free memory with correct free function

[Stefan Berger]

On 5/3/21 12:55 PM, Peter Maydell wrote:
> tpm_util_migration_start_qemu() allocates memory with g_strdup_printf()
> but frees it with free() rather than g_free(), which provokes Coverity
> complaints (CID 1432379, 1432350). Use the correct free function.
>
> Fixes: Coverity CID 1432379, CID 1432350
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> With newer glib (2.46 and up) g_free() is guaranteed to be the same
> as free(), but matching things up is neater anyway.
> ---
>   tests/qtest/tpm-util.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/tests/qtest/tpm-util.c b/tests/qtest/tpm-util.c
> index b70cc32d600..3a40ff3f96c 100644
> --- a/tests/qtest/tpm-util.c
> +++ b/tests/qtest/tpm-util.c
> @@ -289,6 +289,6 @@ void tpm_util_migration_start_qemu(QTestState **src_qemu,
>
>       *dst_qemu = qtest_init(dst_qemu_args);
>
> -    free(src_qemu_args);
> -    free(dst_qemu_args);
> +    g_free(src_qemu_args);
> +    g_free(dst_qemu_args);
>   }


Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>

Re: [PATCH 1/3] tests/qtest/tpm-util.c: Free memory with correct free function

[Alex Bennée]

Peter Maydell <peter.maydell@linaro.org> writes:

> tpm_util_migration_start_qemu() allocates memory with g_strdup_printf()
> but frees it with free() rather than g_free(), which provokes Coverity
> complaints (CID 1432379, 1432350). Use the correct free function.
>
> Fixes: Coverity CID 1432379, CID 1432350
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> With newer glib (2.46 and up) g_free() is guaranteed to be the same
> as free(), but matching things up is neater anyway.
> ---
>  tests/qtest/tpm-util.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/tests/qtest/tpm-util.c b/tests/qtest/tpm-util.c
> index b70cc32d600..3a40ff3f96c 100644
> --- a/tests/qtest/tpm-util.c
> +++ b/tests/qtest/tpm-util.c
> @@ -289,6 +289,6 @@ void tpm_util_migration_start_qemu(QTestState **src_qemu,
>  
>      *dst_qemu = qtest_init(dst_qemu_args);
>  
> -    free(src_qemu_args);
> -    free(dst_qemu_args);
> +    g_free(src_qemu_args);
> +    g_free(dst_qemu_args);
>  }

The more modern approach would be to use g_autofree but this works given
we aren't dealing with multiple exits:

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée

[PATCH 2/3] tests/qtest/rtc-test: Remove pointless NULL check

[Peter Maydell]

In rtc-test.c we know that s is non-NULL because qtest_start()
will return a non-NULL value, and we assume this when we
pass s to qtest_irq_intercept_in(). So we can drop the
initial assignment of NULL and the "if (s)" condition at
the end of the function.

Fixes: Coverity CID 1432353
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 tests/qtest/rtc-test.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/tests/qtest/rtc-test.c b/tests/qtest/rtc-test.c
index 402ce2c6090..8126ab1bdb8 100644
--- a/tests/qtest/rtc-test.c
+++ b/tests/qtest/rtc-test.c
@@ -686,7 +686,7 @@ static void periodic_timer(void)
 
 int main(int argc, char **argv)
 {
-    QTestState *s = NULL;
+    QTestState *s;
     int ret;
 
     g_test_init(&argc, &argv, NULL);
@@ -712,9 +712,7 @@ int main(int argc, char **argv)
 
     ret = g_test_run();
 
-    if (s) {
-        qtest_quit(s);
-    }
+    qtest_quit(s);
 
     return ret;
 }
-- 
2.20.1

Re: [PATCH 2/3] tests/qtest/rtc-test: Remove pointless NULL check

[Thomas Huth]

On 03/05/2021 18.55, Peter Maydell wrote:
> In rtc-test.c we know that s is non-NULL because qtest_start()
> will return a non-NULL value, and we assume this when we
> pass s to qtest_irq_intercept_in(). So we can drop the
> initial assignment of NULL and the "if (s)" condition at
> the end of the function.
> 
> Fixes: Coverity CID 1432353
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>   tests/qtest/rtc-test.c | 6 ++----
>   1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/tests/qtest/rtc-test.c b/tests/qtest/rtc-test.c
> index 402ce2c6090..8126ab1bdb8 100644
> --- a/tests/qtest/rtc-test.c
> +++ b/tests/qtest/rtc-test.c
> @@ -686,7 +686,7 @@ static void periodic_timer(void)
>   
>   int main(int argc, char **argv)
>   {
> -    QTestState *s = NULL;
> +    QTestState *s;
>       int ret;
>   
>       g_test_init(&argc, &argv, NULL);
> @@ -712,9 +712,7 @@ int main(int argc, char **argv)
>   
>       ret = g_test_run();
>   
> -    if (s) {
> -        qtest_quit(s);
> -    }
> +    qtest_quit(s);
>   
>       return ret;
>   }
> 

Reviewed-by: Thomas Huth <thuth@redhat.com>

Re: [PATCH 2/3] tests/qtest/rtc-test: Remove pointless NULL check

[Alex Bennée]

Peter Maydell <peter.maydell@linaro.org> writes:

> In rtc-test.c we know that s is non-NULL because qtest_start()
> will return a non-NULL value, and we assume this when we
> pass s to qtest_irq_intercept_in(). So we can drop the
> initial assignment of NULL and the "if (s)" condition at
> the end of the function.
>
> Fixes: Coverity CID 1432353
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée

[PATCH 3/3] tests: Avoid side effects inside g_assert() arguments

[Peter Maydell]

For us, assertions are always enabled, but side-effect expressions
inside the argument to g_assert() are bad style anyway. Fix three
occurrences in IPMI related tests, which will silence some Coverity
nits.

Fixes: CID 1432322, CID 1432287, CID 1432291
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 tests/qtest/ipmi-bt-test.c  | 6 ++++--
 tests/qtest/ipmi-kcs-test.c | 3 ++-
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/tests/qtest/ipmi-bt-test.c b/tests/qtest/ipmi-bt-test.c
index a42207d416f..8492f02a9c3 100644
--- a/tests/qtest/ipmi-bt-test.c
+++ b/tests/qtest/ipmi-bt-test.c
@@ -98,7 +98,8 @@ static void bt_wait_b_busy(void)
 {
     unsigned int count = 1000;
     while (IPMI_BT_CTLREG_GET_B_BUSY() != 0) {
-        g_assert(--count != 0);
+        --count;
+        g_assert(count != 0);
         usleep(100);
     }
 }
@@ -107,7 +108,8 @@ static void bt_wait_b2h_atn(void)
 {
     unsigned int count = 1000;
     while (IPMI_BT_CTLREG_GET_B2H_ATN() == 0) {
-        g_assert(--count != 0);
+        --count;
+        g_assert(count != 0);
         usleep(100);
     }
 }
diff --git a/tests/qtest/ipmi-kcs-test.c b/tests/qtest/ipmi-kcs-test.c
index fc0a918c8d1..afc24dd3e46 100644
--- a/tests/qtest/ipmi-kcs-test.c
+++ b/tests/qtest/ipmi-kcs-test.c
@@ -73,7 +73,8 @@ static void kcs_wait_ibf(void)
 {
     unsigned int count = 1000;
     while (IPMI_KCS_CMDREG_GET_IBF() != 0) {
-        g_assert(--count != 0);
+        --count;
+        g_assert(count != 0);
     }
 }
 
-- 
2.20.1

Re: [PATCH 3/3] tests: Avoid side effects inside g_assert() arguments

[Philippe Mathieu-Daudé]

On 5/3/21 6:55 PM, Peter Maydell wrote:
> For us, assertions are always enabled, but side-effect expressions
> inside the argument to g_assert() are bad style anyway. Fix three
> occurrences in IPMI related tests, which will silence some Coverity
> nits.
> 
> Fixes: CID 1432322, CID 1432287, CID 1432291
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>  tests/qtest/ipmi-bt-test.c  | 6 ++++--
>  tests/qtest/ipmi-kcs-test.c | 3 ++-
>  2 files changed, 6 insertions(+), 3 deletions(-)

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Re: [PATCH 3/3] tests: Avoid side effects inside g_assert() arguments

[Thomas Huth]

On 03/05/2021 18.55, Peter Maydell wrote:
> For us, assertions are always enabled, but side-effect expressions
> inside the argument to g_assert() are bad style anyway. Fix three
> occurrences in IPMI related tests, which will silence some Coverity
> nits.
> 
> Fixes: CID 1432322, CID 1432287, CID 1432291
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>   tests/qtest/ipmi-bt-test.c  | 6 ++++--
>   tests/qtest/ipmi-kcs-test.c | 3 ++-
>   2 files changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/tests/qtest/ipmi-bt-test.c b/tests/qtest/ipmi-bt-test.c
> index a42207d416f..8492f02a9c3 100644
> --- a/tests/qtest/ipmi-bt-test.c
> +++ b/tests/qtest/ipmi-bt-test.c
> @@ -98,7 +98,8 @@ static void bt_wait_b_busy(void)
>   {
>       unsigned int count = 1000;
>       while (IPMI_BT_CTLREG_GET_B_BUSY() != 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);
>           usleep(100);
>       }
>   }
> @@ -107,7 +108,8 @@ static void bt_wait_b2h_atn(void)
>   {
>       unsigned int count = 1000;
>       while (IPMI_BT_CTLREG_GET_B2H_ATN() == 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);
>           usleep(100);
>       }
>   }
> diff --git a/tests/qtest/ipmi-kcs-test.c b/tests/qtest/ipmi-kcs-test.c
> index fc0a918c8d1..afc24dd3e46 100644
> --- a/tests/qtest/ipmi-kcs-test.c
> +++ b/tests/qtest/ipmi-kcs-test.c
> @@ -73,7 +73,8 @@ static void kcs_wait_ibf(void)
>   {
>       unsigned int count = 1000;
>       while (IPMI_KCS_CMDREG_GET_IBF() != 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);
>       }
>   }

According to 
https://developer.gnome.org/glib/unstable/glib-Testing.html#g-assert 
g_assert() should be avoided in unit tests and g_assert_true() should be 
used instead. So I think it might be nicer to use g_assert_true() here?

  Thomas

Re: [PATCH 3/3] tests: Avoid side effects inside g_assert() arguments

[Peter Maydell]

On Tue, 4 May 2021 at 08:18, Thomas Huth <thuth@redhat.com> wrote:
>
> On 03/05/2021 18.55, Peter Maydell wrote:
> > For us, assertions are always enabled, but side-effect expressions
> > inside the argument to g_assert() are bad style anyway. Fix three
> > occurrences in IPMI related tests, which will silence some Coverity
> > nits.
> >
> > Fixes: CID 1432322, CID 1432287, CID 1432291
> > Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

> > diff --git a/tests/qtest/ipmi-kcs-test.c b/tests/qtest/ipmi-kcs-test.c
> > index fc0a918c8d1..afc24dd3e46 100644
> > --- a/tests/qtest/ipmi-kcs-test.c
> > +++ b/tests/qtest/ipmi-kcs-test.c
> > @@ -73,7 +73,8 @@ static void kcs_wait_ibf(void)
> >   {
> >       unsigned int count = 1000;
> >       while (IPMI_KCS_CMDREG_GET_IBF() != 0) {
> > -        g_assert(--count != 0);
> > +        --count;
> > +        g_assert(count != 0);
> >       }
> >   }
>
> According to
> https://developer.gnome.org/glib/unstable/glib-Testing.html#g-assert
> g_assert() should be avoided in unit tests and g_assert_true() should be
> used instead. So I think it might be nicer to use g_assert_true() here?

That probably depends on what we decide about whether we want to
use glib-testing-style "assert but this might not stop execution" in our
tests: see this thread:
https://lore.kernel.org/qemu-devel/CAFEAcA9juOChqrh5orybJQwpQsyEZ5z3Dvmy7fjX0DW4Nbgmrg@mail.gmail.com/
(I should have cc'd you and Laurent on that as qtest maintainers; sorry.)

-- PMM

Re: [PATCH 3/3] tests: Avoid side effects inside g_assert() arguments

[Alex Bennée]

Peter Maydell <peter.maydell@linaro.org> writes:

> For us, assertions are always enabled, but side-effect expressions
> inside the argument to g_assert() are bad style anyway. Fix three
> occurrences in IPMI related tests, which will silence some Coverity
> nits.
>
> Fixes: CID 1432322, CID 1432287, CID 1432291
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>  tests/qtest/ipmi-bt-test.c  | 6 ++++--
>  tests/qtest/ipmi-kcs-test.c | 3 ++-
>  2 files changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/tests/qtest/ipmi-bt-test.c b/tests/qtest/ipmi-bt-test.c
> index a42207d416f..8492f02a9c3 100644
> --- a/tests/qtest/ipmi-bt-test.c
> +++ b/tests/qtest/ipmi-bt-test.c
> @@ -98,7 +98,8 @@ static void bt_wait_b_busy(void)
>  {
>      unsigned int count = 1000;
>      while (IPMI_BT_CTLREG_GET_B_BUSY() != 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);

This does seem a little weird - we are not asserting an interface
violation just that the read should have cleared in 1000 * 100 usec. If
it doesn't is that really a theoretically impossible situation or just
an example of a failed test.

That said looking at how deeply buried in the test these helpers are an
assert is probably better than a convoluted attempt to return out and
exit the test with a failure.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>


>          usleep(100);
>      }
>  }
> @@ -107,7 +108,8 @@ static void bt_wait_b2h_atn(void)
>  {
>      unsigned int count = 1000;
>      while (IPMI_BT_CTLREG_GET_B2H_ATN() == 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);
>          usleep(100);
>      }
>  }
> diff --git a/tests/qtest/ipmi-kcs-test.c b/tests/qtest/ipmi-kcs-test.c
> index fc0a918c8d1..afc24dd3e46 100644
> --- a/tests/qtest/ipmi-kcs-test.c
> +++ b/tests/qtest/ipmi-kcs-test.c
> @@ -73,7 +73,8 @@ static void kcs_wait_ibf(void)
>  {
>      unsigned int count = 1000;
>      while (IPMI_KCS_CMDREG_GET_IBF() != 0) {
> -        g_assert(--count != 0);
> +        --count;
> +        g_assert(count != 0);
>      }
>  }


-- 
Alex Bennée