[igt-dev] [PATCH i-g-t v5 2/5] lib/igt_pm: Add lib func to get lpsp capability

[Anshuman Gupta]

On 2020-04-14 at 21:56:47 +0530, Manna, Animesh wrote:
> 
> On 09-04-2020 11:09, Anshuman Gupta wrote:
> >This lib function evaluate the lpsp capability from
> >the connector specific debugfs attribute i915_lpsp_info.
> >
> >v2:
> >- changed the lib function prefix igt_output_is_lpsp_capable
> >   to i915_output_is_lpsp_capable. [Martin]
> >v3:
> >- early return for the connector which doesn't support
> >   lpsp on any platform.
> >v4:
> >- debugfs entry changed from i915_lpsp_info to i915_lpsp_capability.
> >
> >Signed-off-by: Anshuman Gupta <anshuman.gupta at intel.com>
> >---
> >  lib/igt_pm.c | 37 +++++++++++++++++++++++++++++++++++++
> >  lib/igt_pm.h |  1 +
> >  2 files changed, 38 insertions(+)
> >
> >diff --git a/lib/igt_pm.c b/lib/igt_pm.c
> >index 9d441e1b..3418b69a 100644
> >--- a/lib/igt_pm.c
> >+++ b/lib/igt_pm.c
> >@@ -37,6 +37,7 @@
> >  #include <dirent.h>
> >  #include "drmtest.h"
> >+#include "igt_kms.h"
> >  #include "igt_pm.h"
> >  #include "igt_aux.h"
> >  #include "igt_sysfs.h"
> >@@ -827,3 +828,39 @@ bool igt_pm_pc8_plus_residencies_enabled(int msr_fd)
> >  	return true;
> >  }
> >+
> >+/**
> >+ * i915_output_is_lpsp_capable:
> >+ * @drm_fd: fd to drm device
> >+ * @output: igt output for which lpsp capability need to be evaluated
> >+ * Check lpsp capability for a given output.
> >+ *
> >+ * Returns:
> >+ * True if given output is lpsp capable otherwise false.
> >+ */
> >+bool i915_output_is_lpsp_capable(int drm_fd, igt_output_t *output)
> >+{
> >+	drmModeConnectorPtr c = output->config.connector;
> >+	char buf[256];
> >+	int fd, len;
> >+
> >+	/* only eDP/DP/DSI/HDMI can support LPSP */
> >+	if (c->connector_type != DRM_MODE_CONNECTOR_eDP &&
> >+	    c->connector_type != DRM_MODE_CONNECTOR_DSI &&
> >+	    c->connector_type != DRM_MODE_CONNECTOR_DisplayPort &&
> >+	    c->connector_type != DRM_MODE_CONNECTOR_HDMIA &&
> >+	    c->connector_type != DRM_MODE_CONNECTOR_HDMIB)
> >+		return false;
> 
> We can avoid check for Connector type as already taken care in kernel... rt?
Thanks for review animesh.
kernel will not create the debugfs for legacy connector in that case there will
not be any entry, as there will not be any entry later this function assert on
the debugfs entry, which will fail the test.
That is why connector check is there to return the capability false, when
there is no entry. 
Thanks,
Anshuman Gupta.
> 
> Regards,
> Animesh
> 
> >+
> >+	fd = igt_debugfs_connector_dir(drm_fd, output->name, O_RDONLY);
> >+	igt_require(fd >= 0);
> >+	len = igt_debugfs_simple_read(fd, "i915_lpsp_capability",
> >+				      buf, sizeof(buf));
> >+
> >+	if (len < 0)
> >+		igt_assert_eq(len, -ENODEV);
> >+
> >+	close(fd);
> >+
> >+	return strstr(buf, "LPSP: capable");
> >+}
> >diff --git a/lib/igt_pm.h b/lib/igt_pm.h
> >index 5e438452..162d3ca3 100644
> >--- a/lib/igt_pm.h
> >+++ b/lib/igt_pm.h
> >@@ -53,5 +53,6 @@ enum igt_runtime_pm_status igt_get_runtime_pm_status(void);
> >  bool igt_wait_for_pm_status(enum igt_runtime_pm_status status);
> >  bool igt_pm_dmc_loaded(int debugfs);
> >  bool igt_pm_pc8_plus_residencies_enabled(int msr_fd);
> >+bool i915_output_is_lpsp_capable(int drm_fd, igt_output_t *output);
> >  #endif /* IGT_PM_H */