Most cosmic ray particles observed derive from the explosions of massive stars. Massive stars from slightly above about 10M⊙ explode as supernovae via a mechanism which we do not know yet: two not mutually exclusive main ideas are an explosion driven by neutrinos, or the magneto-rotational mechanism, in which the magnetic field acts like a conveyor-belt to transport energy outwards for an explosion. Massive stars above about 25M⊙, depending on their heavy element abundance, commonly produce stellar black holes in their supernova explosions. When two such black holes find themselves in a tight binary system they finally merge in a gigantic emission of gravitational waves, events that have now been detected. The radio interferometric data demonstrate that all of these stars have powerful magnetic winds. After an introduction (Section 1) we introduce the basic concept (Section 2): Cosmic rays from exploding massive stars with winds always show two cosmic ray components at the same time: (i) the weaker polar cap component only produced by Diffusive Shock Acceleration, showing a relatively flat spectrum, and cut-off at the knee, and (ii) the stronger 4π component, which is produced by a combination of Stochastic Shock Drift Acceleration and Diffusive Shock Acceleration, with a down-turn to a steeper power-law spectrum at the knee, and a final cut-off at the ankle. In Section 3 we use the Alpha Magnetic Spectrometer (AMS) data to differentiate these two cosmic ray spectral components; these two cosmic ray components excite magnetic irregularity spectra in the plasma, and the ensuing secondary spectra can explain anti-protons, lower energy positrons, and other secondary particles. Cosmic ray electrons of the polar cap component interact with the surrounding photon field to produce positrons by triplet pair production, and in this manner may explain the higher energy positron AMS data. In Section 4 we test this paradigm with a theory of injection based on a combined effect of first and second ionization potential; this reproduces the ratio of cosmic ray source abundances to source material abundances. We can interpret the abundance data using the relation of the total number of ions enhanced by Q02A+2/3, where Q0 is the initial degree of ionization, and A is the mass number. This interpretation implies the high temperature as observed in the winds of blue super-giant stars; it also requires that cosmic ray injection happens in the shock travelling through such a wind. Most injection happens at the largest radii before slowing down due to interaction with the environment. In Section 5 we interpret the compact radio source 41.9 + 58 in the starburst galaxy M82 as a recent binary black hole merger, with an accompanying gamma ray burst. The tell-tale observational sign is the conical cleaning sweep of the relativistic jet during the merger, observed as an open cone with very low radio emission. This can also explain the Ultra High Energy Cosmic Ray (UHECR) data in the Northern sky. Thus, by studying the cosmic ray particles, their abundances at knee energies, and their spectra, we can learn about what drives these stars to produce the observed cosmic rays.