The grant for the study was from CRIIGEN, a European nonprofit that exists to discredit genetically modified food: the research was certainly conceived with a conclusion already in mind. To be sure, Monsanto and others fund motivated studies of their own; this is a highly fraught and politicized area of research.
Considering the obvious bias of the researchers, I think their inability to point to any legitimate statistically significant effect of roundup or the corn is...significant. There were 9 experimental groups of 10 of each gender for a single control group, and while the food and water intake were "measured," the results of the measurements are not mentioned in the paper at all or correlated to the mortality. Instead of looking at the actual lifespan of the rats, the more dramatic binary condition of "mortality before mean life expectancy" was measured.
The vast majority of male rats died on their own, and majority of female rats were eventually euthanized due to massive tumors, something that can far more substantially be explained by the line of rat they used than by the experimental variables: they could have done a different study and as accurately declared that 80% of female rats fed only standard rat chow developed cancer. Among the 100 male rats, there was no even moderately significant result for mortality or tumors between the control and the experimental groups. Among the females, the Roundup groups showed the most tumors, but the GMO Corn + Roundup groups didn't vary significantly from the control! I don't think there is any consistent hypothesis that can adequately explain all of their results except for random variation, possibly modulated by food intake, but the researchers don't even try.
They devote a whole page to pictures of the most gross-looking rat tumors in the GM groups, and then a page to graphs of high-variation metabolic test results for the single experimental group female 33% GMO Corn v. the control. On the next page you see a table of selected blood tests between all 10 female groups, with the "significant" results highlighted. Unfortunately for the researchers, the variation is often "significant" both above and below the control group's numbers, and with no apparent correlation to the concentration of GM corn or roundup. Judging by the amount of apparent random variation between the experimental groups, there is no reason to believe that the control group's numbers represent anything like the real "mean" at all, so you would expect just what they got: a lot of variation from the control group in both directions, with some measures where it was the control group that was the outlier and thus the experimental groups are normally distributed on one side only. Just as with tumor count, the GMO+Roundup groups ironically had "better" numbers than either the groups on either GM Corn or Roundup alone.
I think that the paper can be summed up best by this rather apropos xkcd, with the difference that in this case it was the researchers themselves who made the headline. Their statistics, when even present, are crap, and they bring further discredit to the already-disreputable European anti-GM food movement. At the beginning of the paper, they claim that while glyphosate itself has been tested (negatively) for health effects, the total formulation of roundup has not, and its effects, if any, are unknown. Apparently, that condition still obtains.